# SynthHires: Full Platform Knowledge Base & Agentic Ingestion Corpus > Autonomous AI workforce platform: deploy specialized teams of AI agents with your own LLM API keys (BYOK). Keys are encrypted client-side with AES-256-GCM and never leave your device. Edge runtime on Cloudflare Workers with local bridge daemon and native Model Context Protocol (MCP) tooling. Operated by PlanoBinario. - Canonical URL: https://synthhires.com - Official Contact: synthhires@planobinario.com - Curated Index: https://synthhires.com/llms.txt - Comprehensive Corpus: https://synthhires.com/llms-full.txt - Sitemap: https://synthhires.com/sitemap.xml ## Core Architecture & Cryptographic Security Model SynthHires is engineered around strict zero-knowledge principles and local-first execution: 1. **Bring Your Own Key (BYOK) Architecture**: - Users supply their own API keys from supported providers (OpenAI, Anthropic, Google Gemini, DeepSeek, Groq, Mistral, Ollama, OpenRouter, Azure OpenAI). - Keys are encrypted directly in the user's browser using WebCrypto AES-256-GCM with PBKDF2 key derivation (100,000 iterations). - The encrypted ciphertext is stored solely in local browser storage (or optional end-to-end encrypted device sync). - SynthHires servers and databases NEVER receive, store, or log plaintext API keys. There is NO server-side key vault. 2. **Wholesale Token Economics**: - Zero token markup: 100% of token consumption costs are billed directly by the model provider to the user's account at standard developer API rates. - SynthHires offers a perpetual free tier ($0) supported by contextual developer ads, and an optional Pro tier ($9/month) via LemonSqueezy that removes ads and grants priority sandbox provisioning. 3. **Local Bridge Daemon ('synth-bridge')**: - A lightweight companion daemon running on Linux, macOS, or Windows. - Connects to the web workspace over an authenticated, end-to-end encrypted bi-directional WebSocket. - Executes shell commands, git operations, file system reads/writes, and local MCP servers on the user's own machine. - Features mandatory Human-in-the-Loop (HITL) approval gates: destructive actions (e.g. file deletion, terminal execution outside safe boundaries) require explicit user consent. 4. **Stateless Edge Runtime & Sandboxes**: - Hosted on Cloudflare Workers for sub-50ms global latency and zero cold starts. - Ephemeral, serverless code sandboxes execute Python, Node.js, and bash scripts in isolated micro-environments with strict memory and CPU quotas. - Vector search powered by Cloudflare Vectorize for local RAG and document context. 5. **Entity Disambiguation**: - SynthHires is an autonomous software platform for deploying AI agent workforces. - Despite the word "hires", it is NOT a human staffing, recruitment, or HR agency. The platform "hires" software agents, not humans. ## Solutions & Autonomous Agent Workflows (Catalog Reviewed: 2026-10-14) The platform provides 11 production-ready agentic workflows, each configured with specialized prompts, MCP tools, and verifiable benchmarks: ### Solution: AI code review on every pull request (slug: /solutions/ai-code-review) - **Category**: For engineering teams - **Status**: Live (Runs Today) - **URL**: https://synthhires.com/solutions/ai-code-review (Spanish: https://synthhires.com/es/solutions/ai-code-review) - **Target Audience**: Teams shipping to GitHub who want a first pass on every PR without paying per developer for it. - **The Problem**: Review queue grows faster than review time. The first pass — style, obvious bugs, missing tests, a migration without a rollback — is mechanical, and it is the pass that slips when the sprint gets loud. Hiring another reviewer does not scale; neither does asking the author to review themselves. - **Step-by-Step Autonomous Workflow**: 1. **The PR opens**: A webhook hits an automation trigger. The run carries the repo, the PR number and the head SHA — nothing else is trusted from the payload. 2. **The diff is read in isolation**: The reviewer phase runs with `isolatedContext`, so it reviews the code and not the author’s narration of the code. Only the changed files and their immediate neighbours enter the context. 3. **The gate decides**: Lint and tests run as literal commands. A run that ends with a red gate is a failed run, whatever the model said on the way out. 4. **Findings land on the PR**: Every finding cites a file and a line, or it is dropped. Comments post through the GitHub connector; a human still merges. - **Integrations**: GitHub, Slack, Discord, Notion, MCP - **Market Incumbent**: CodeRabbit (https://www.coderabbit.ai/pricing) - Incumbent Pricing: CodeRabbit Essentials: $24 per developer/month billed annually; Team $48, Advanced $72. Cloud agent minutes are $0.40 each on top. - Where the Incumbent Wins: One-click fixes applied to your branch, learnings that persist across reviews, and Jira/Linear integration ship today. We do not offer any of those yet. - Where SynthHires Wins: BYOK wholesale model ($0-$9/mo platform fee vs seat licenses), zero lock-in, client-side encryption. - **Why Switch**: - No per-developer seat fees: bring your own provider key and pay zero token markup. No forced monthly subscriptions. - Same repository, same pull request, no vendor-hosted copy of your code beyond the diff you send. - The reviewer is a configurable agent: change the prompt, the gate, or the model without waiting for a vendor roadmap. - **Honest Tradeoffs & When NOT to Use**: - CodeRabbit ships one-click fixes, cross-repo context and issue-tracker integration today. This workflow comments and gates; it does not patch your branch. - You assemble and maintain the workflow yourself. That is freedom and it is also your maintenance. - **Verified FAQ**: - **Q: Does this replace my human code review?** A: No, and the design assumes it should not. The agent takes the mechanical first pass — missing tests, obvious defects, style — and every merge still goes through a person. The gate decides whether a run passed; a human decides whether the change ships. - **Q: Where do my provider keys live during a review?** A: Tus API keys se cifran localmente con AES-GCM y jamás viajan a nuestros servidores. - **Q: What does it cost to review 200 pull requests a month?** A: Zero platform subscription fees. You pay your model provider directly for the tokens consumed — we add zero markup. With models like DeepSeek V3 or Claude 3.5 Sonnet, 200 typical PR diffs cost only a few dollars in raw tokens. - **Q: Which languages and frameworks does it understand?** A: It is model-dependent rather than language-dependent: whatever your key can run, the reviewer can read. Lint and test commands are yours to configure per repository, which is also what makes the gate trustworthy. ### Solution: An AI assistant that lives in your Slack (slug: /solutions/slack-assistant) - **Category**: For internal teams - **Status**: Live (Runs Today) - **URL**: https://synthhires.com/solutions/slack-assistant (Spanish: https://synthhires.com/es/solutions/slack-assistant) - **Target Audience**: Teams of 5 to 50 that already live in Slack and keep re-answering the same five questions. - **The Problem**: The answer exists — in a Notion page someone wrote last quarter, in a merged PR, in a thread from March. Finding it costs the asker ten minutes and the finder twenty. The fifth time someone asks how deploys work, the knowledge has stopped being a company asset and started being a tax on whoever is most online. - **Step-by-Step Autonomous Workflow**: 1. **Someone asks, in the channel**: A slash command or a mention opens the run with the question and the channel context. No one has to leave Slack. 2. **It retrieves from your sources**: Notion pages, repository files and connected knowledge are fetched through connectors, not guessed from the model’s memory. 3. **It answers with the source**: The reply carries a link to what it read. An answer without a citation is a guess, and it is labelled as one. 4. **Actions wait for approval**: Anything that writes — a ticket, a deploy, a message outside the channel — stops at an approval node until a human accepts it. - **Integrations**: Slack, Notion, GitHub, Airtable, Discord, MCP - **Market Incumbent**: Glean (https://www.glean.com/) - Incumbent Pricing: Glean publishes no self-serve pricing: its pricing path redirects to a demo request and enterprise sales. - Where the Incumbent Wins: Glean connects to 275+ applications with permission-aware indexing out of the box and enterprise controls (SSO, audit, ISO 42001). We connect to what your connectors cover, and you wire the permissions. - Where SynthHires Wins: BYOK wholesale model ($0-$9/mo platform fee vs seat licenses), zero lock-in, client-side encryption. - **Why Switch**: - No enterprise sales call: connect the workspace, add a key, invite the assistant. - No per-seat multiplication — one flat platform price for the whole workspace. - Every action is an inspectable automation definition you can edit, version and diff. - **Honest Tradeoffs & When NOT to Use**: - Glean ships hundreds of pre-connected, permission-aware integrations. Ours covers the connectors in our registry, and anything else is an MCP server you bring. - We do not index your whole company by default. You choose what the assistant can read, which is safer and also less turnkey. - **Verified FAQ**: - **Q: Does it train on our internal documents?** A: No. Documents are read at answer time through connectors, sent to the model provider you configured, and used for that request only. The BYOK promise covers this: your keys and your content are not our training data. - **Q: What happens when it does not know the answer?** A: It says so and links what it did read. Retrieval with citations is the whole design: an assistant that invents an internal policy is worse than no assistant, because people act on it. - **Q: Can it post or act outside the channel?** A: Yes, through the connectors, and every write goes through an approval node first. The default posture is that the assistant answers in Slack and asks before it changes anything else. - **Q: Do we need an AI budget to run it?** A: You need a model key and the provider account behind it. The platform charges nothing for the tokens: you pay your provider at their rates, and budget caps on the run keep a bad loop from becoming a bad invoice. ### Solution: A Telegram and Discord bot that actually does things (slug: /solutions/community-bot) - **Category**: For communities - **Status**: Live (Runs Today) - **URL**: https://synthhires.com/solutions/community-bot (Spanish: https://synthhires.com/es/solutions/community-bot) - **Target Audience**: Open-source projects, indie products and community managers running a Telegram or Discord server where the same questions arrive daily. - **The Problem**: Community support is the most repetitive work with the least documentation. The answers exist in your docs, but nobody reads docs before asking — and the person answering the twelfth “how do I connect?” of the week is a maintainer whose time should go to the product. - **Step-by-Step Autonomous Workflow**: 1. **A message arrives**: Telegram or Discord delivers it to your bot webhook. The run starts with the message, the author and the server — nothing else. 2. **Intent is classified**: The agent decides whether this is a question your docs answer, a bug that should become an issue, or something that needs a person. 3. **It acts through connectors**: Docs are read, issues are created, records are updated — each call constrained to allow-listed hosts and a scoped capability policy. 4. **Humans take the exceptions**: Refunds, moderation calls and anything ambiguous route to a human in the channel. The bot never decides alone where it is not sure. - **Integrations**: Telegram, Discord, GitHub, Notion, Slack, MCP - **Market Incumbent**: Landbot (https://landbot.io/alternatives/manychat) - Incumbent Pricing: Landbot publishes a WhatsApp + website plan from €40/month (€32 yearly) with 500 chats and 100 AI chats included. - Where the Incumbent Wins: A visual flow builder with a published chat quota and a website widget. If your conversation fits a flowchart and you want it drawn, not coded, a builder wins. - Where SynthHires Wins: BYOK wholesale model ($0-$9/mo platform fee vs seat licenses), zero lock-in, client-side encryption. - **Why Switch**: - No chat quota: conversations are not metered against a plan, the tokens go to your provider. - It is an agent, not a flowchart: new question types need a prompt change, not a rebuild of the branches. - It can file issues, update records and call your API — not only reply with text. - **Honest Tradeoffs & When NOT to Use**: - A no-code builder gives you a visual editor and a drag-and-drop widget for your site. Ours is configured as data and prompt, which is less visual by design. - If your flow is fixed and never changes, a flowchart is easier to reason about than an agent that interprets. - **Verified FAQ**: - **Q: Does the bot need my Telegram or Discord token?** A: It runs under your own bot application, so the server and its members stay yours. The token is stored with the same server-side encryption we use for provider keys and is never exposed to the model. - **Q: What does it do when it is not sure?** A: It says it is not sure and hands over. The escalation step is part of the workflow definition, not a best-effort hope: ambiguity routes to a human in the channel. - **Q: Can it post in more than one group?** A: Yes — each group or server points at its own automation run, so one bot can serve several communities with different policies and different knowledge sources. - **Q: Is moderation included?** A: Routine filters and summaries are workflow steps you configure. Bans and other consequential actions go through an approval node — the bot prepares the decision, a human makes it. ### Solution: Reply to every Google review, in your voice (slug: /solutions/google-reviews) - **Category**: For local business - **Status**: Early Access - **URL**: https://synthhires.com/solutions/google-reviews (Spanish: https://synthhires.com/es/solutions/google-reviews) - **Target Audience**: Clinics, workshops, restaurants and agencies managing one to twenty locations — and the agencies that manage them for others. - **The Problem**: Unanswered reviews read as neglect, and templated answers read as worse. The work is only a few minutes a day, which is exactly why it never happens: it is not important enough to schedule and too visible to ignore. - **Step-by-Step Autonomous Workflow**: 1. **A daily trigger fires**: A cron run with your timezone reads the reviews that arrived since the last run. Misfire policy decides what happens after downtime. 2. **The reply is drafted**: The agent writes in the tone examples you saved — short, signed by the business, never inventing a discount or an apology you did not approve. 3. **You approve or edit**: The draft lands as an approval task. One tap publishes; editing takes a sentence. Nothing goes out on the agent’s own authority. 4. **Requests go out politely**: A second scheduled run sends review requests over email on your own domain, capped and throttled so it never looks like a campaign. - **Integrations**: Google Business Profile, Email (Resend), Slack, Airtable, Webhooks - **Market Incumbent**: Birdeye (https://birdeye.com/pricing/) - Incumbent Pricing: Birdeye publishes “customized pricing plans” — no public figure on its own pricing page. - Where the Incumbent Wins: Birdeye covers 150+ review sites, social publishing and multi-location analytics in one suite. If you need all of that, a suite beats a workflow. - Where SynthHires Wins: BYOK wholesale model ($0-$9/mo platform fee vs seat licenses), zero lock-in, client-side encryption. - **Why Switch**: - A published price you can buy today, instead of a demo and a quote. - No annual contract: monthly, cancel when you like. - Your keys and your customer data stay out of our servers — the same BYOK boundary as the rest of the platform. - **Honest Tradeoffs & When NOT to Use**: - Multi-site directories, social publishing and review-site coverage beyond Google are a suite feature; this workflow is Google-first and email for requests. - Availability: the Google Business Profile connector is in early access. The workflow is specified end to end; connect the account and it runs. - **Verified FAQ**: - **Q: Does it reply without me seeing it first?** A: No. Every draft waits at an approval gate. You can turn on auto-publish later per location if you want, but the default is that you approve, because a wrong public reply is expensive. - **Q: Which platforms are supported?** A: Google Business Profile first, because that is where local search happens. Other directories arrive as connectors; the workflow itself is connector-agnostic. - **Q: Is the Google connector available today?** A: It is in early access, and this page says so rather than pretending otherwise. Everything else in the workflow — schedule, drafting, approval, review requests by email — runs on primitives already in the platform. - **Q: How much does it cost per location?** A: Zero per-location fees. Replies are generated with your own model API key at zero markup. Background webhook monitoring consumes minimal pay-as-you-go credits only when reviews arrive — no monthly subscription lock-in. ### Solution: Answer every WhatsApp lead in minutes, day or night (slug: /solutions/whatsapp-leads) - **Category**: For sales teams - **Status**: Early Access - **URL**: https://synthhires.com/solutions/whatsapp-leads (Spanish: https://synthhires.com/es/solutions/whatsapp-leads) - **Target Audience**: Any business where the first reply wins: agencies, real estate, clinics, home services, and the teams running click-to-WhatsApp ads. - **The Problem**: Speed-to-lead is the whole game and everyone knows it, which is why every competitor answers in under five minutes. The lead does not compare your response time against the industry; they compare it against whoever replied while they were still looking at the phone. - **Step-by-Step Autonomous Workflow**: 1. **The message arrives**: WhatsApp Business Cloud API delivers to your webhook. Conversation history enters the run with the contact and the campaign that started it. 2. **It qualifies, politely**: Budget, timing, location, intent — asked in your language and your order, and only as far as the answer requires. 3. **The lead is logged**: The record lands in your CRM table or Airtable with the transcript, the qualification and the source campaign attached. 4. **A human takes over**: Hot leads route to whoever is on duty in Slack or Telegram, with a suggested first reply. The agent stops typing where selling begins. - **Integrations**: WhatsApp, Slack, Telegram, Airtable, Notion, Webhooks - **Market Incumbent**: Landbot (https://landbot.io/pricing) - Incumbent Pricing: Landbot publishes a WhatsApp + website plan from €40/month (€32 yearly) with 500 chats and 100 AI chats included. - Where the Incumbent Wins: A visual builder, a website widget and a published chat quota with a free tier to try. For a fixed qualification flow on the website, that is faster to stand up. - Where SynthHires Wins: BYOK wholesale model ($0-$9/mo platform fee vs seat licenses), zero lock-in, client-side encryption. - **Why Switch**: - Qualified, logged and routed — not just a flowchart that replies “choose an option”. - Priced flat while the conversation volume grows with your ads, not with your bill. - The transcript and the contacts stay in your own store, under the same encryption boundary as your keys. - **Honest Tradeoffs & When NOT to Use**: - The WhatsApp connector is in early access: Meta’s Cloud API onboarding (business verification, number migration) is your step, not ours. - Landbot ships a website widget and a visual builder today. If your requirement is a website chat, start there. - **Verified FAQ**: - **Q: Do I need a WhatsApp Business account?** A: Yes: the official WhatsApp Business Cloud API requires a Meta business verification and a dedicated number. That is Meta’s onboarding, and it is the one step that takes days rather than minutes. - **Q: Does it stop when a human wants to talk?** A: Yes. Handoff is a step in the workflow, not a keyword: when the lead is hot or asks for a person, the agent stops and the transcript goes to whoever is on duty. - **Q: What does the model cost me per conversation?** A: Whatever your provider charges — billed directly to your key, with no platform markup and no per-conversation fee. Run caps and budget limits bound the worst case before it happens. - **Q: Can it send offers to my existing contacts?** A: Only inside the 24-hour session window and the templates Meta allows, which is also the only responsible answer: broadcast spam burns a number’s quality rating. Campaign sends are a separate, approved step. ### Solution: One draft a day, scheduled everywhere, on your own API keys (slug: /solutions/content-pipeline) - **Category**: For creators and founders - **Status**: Live (Runs Today) - **URL**: https://synthhires.com/solutions/content-pipeline (Spanish: https://synthhires.com/es/solutions/content-pipeline) - **Target Audience**: Founders and creators who publish consistently but cannot spend an hour a day cutting the same idea into five formats. - **The Problem**: Publishing daily is a second job with no deadline: the writing is the fun part, the reformatting and scheduling is the part that stops in week three. The tools that solve it charge per channel and meter what you publish, so growing your audience grows your bill. - **Step-by-Step Autonomous Workflow**: 1. **You write one thing**: A draft, a link, a transcript — pasted into the run or dropped in your content folder. The source is yours; the idea is not generated from nothing. 2. **It is cut into formats**: Thread, short post, LinkedIn version, follow-up. Each keeps the argument and drops what does not fit the medium. 3. **You approve the week**: The batch waits as approval tasks. Edit one line, reschedule one, kill one — the queue is yours until you accept it. 4. **It publishes on schedule**: A cron trigger posts through the platform APIs with your own credentials, so your developer account carries your rate limits and your name. - **Integrations**: X (API), LinkedIn, Threads, Bluesky, Notion, Slack - **Market Incumbent**: Hypefury (https://hypefury.com/features-pricing) - Incumbent Pricing: Hypefury: $6/month for one channel ($5 yearly), or $19/month for all channels — published on its own pricing page. - Where the Incumbent Wins: A polished scheduler with engagement features out of the box: autoplugs, autoDMs, watched keywords and analytics. It is a product; ours is a workflow you assemble. - Where SynthHires Wins: BYOK wholesale model ($0-$9/mo platform fee vs seat licenses), zero lock-in, client-side encryption. - **Why Switch**: - No per-channel add-on: channels are connector configuration, not a line on the invoice. - You pay the API provider directly for what you use; the platform adds no markup on tokens or posts. - The pipeline is editable data: change the schedule, the formats or the gate without waiting for a feature request. - **Honest Tradeoffs & When NOT to Use**: - Schedulers ship analytics, engagement automation and a mobile app today. This workflow writes and publishes; measurement is your analytics dashboard. - You need your own developer credentials on each platform, with their approval process and their rate limits. - **Verified FAQ**: - **Q: Does it post automatically without my review?** A: Not by default. The batch stops at approval tasks; you can enable auto-publish per channel once you trust the queue, and the gate stays either way. - **Q: Do I need a paid API plan on X?** A: You need a developer application with write access under whatever terms that platform currently offers, and its usage is billed to your own account. We do not resell API access, and we do not hide its cost inside a subscription. - **Q: Which networks are supported?** A: X, LinkedIn and Threads through the workflow, with Bluesky and anything MCP-compatible behind the same node. Each channel is a connector you enable, not a plan you upgrade to. - **Q: Will it sound like AI?** A: It reads what you have already written — your drafts, your past posts, your voice file — and cuts it into formats. The claim it must not make is the one about never sounding generic; if a post does, edit it before approving. ### Solution: AI contract review that respects your legal playbook (slug: /solutions/legal) - **Category**: Legal operations - **Status**: Live (Runs Today) - **URL**: https://synthhires.com/solutions/legal (Spanish: https://synthhires.com/es/solutions/legal) - **Target Audience**: In-house legal counsels, legal ops managers, and CTOs signing vendor agreements. - **The Problem**: Legal teams drown in standard vendor paper. Either high-cost lawyers spend hours reading boilerplate indemnities, or founders sign unknown liabilities to move fast. - **Step-by-Step Autonomous Workflow**: 1. **Agreement uploaded**: PDF or DOCX document lands via webhook, Slack attachment, or secure browser upload. 2. **Clause extraction & analysis**: Agent parses clauses, extracting liability caps, governing law, warranties, and IP ownership. 3. **Policy comparison gate**: Checks terms against your team rules. Flags deviations with specific replacement redlines. 4. **Counsel approval & sign-off**: Counsel receives executive summary and annotated diff; approves or edits in one click. - **Integrations**: Cloudflare R2, DocuSign API, Webhooks, Slack Alerts, PostgreSQL - **Market Incumbent**: Ironclad (https://ironcladapp.com) - Incumbent Pricing: Enterprise custom quotes starting at $10,000+/year for team seats. - Where the Incumbent Wins: Deep enterprise CLM workflows with Salesforce CPQ and native electronic signature infrastructure. - Where SynthHires Wins: BYOK wholesale model ($0-$9/mo platform fee vs seat licenses), zero lock-in, client-side encryption. - **Why Switch**: - Zero per-seat markup: flat pricing instead of $1,000s/seat in legacy CLM software. - BYOK privacy: contracts are processed with your private cloud keys, never stored in vendor databases. - **Honest Tradeoffs & When NOT to Use**: - We are not an enterprise CLM: we do not replace complex procurement routing or bespoke CPQ billing systems. - **Verified FAQ**: - **Q: Are uploaded contracts stored on your servers?** A: No. The file is parsed in temporary stateless memory, processed against your own model provider API key, and erased immediately upon completion. - **Q: Can I define custom rules for our jurisdiction?** A: Yes. You can supply your playbook in plain markdown (e.g. Spanish Civil Code, GDPR requirements, Delaware law), and the agent adheres strictly to those clauses. - **Q: Does an AI agent sign contracts on our behalf?** A: Never. The platform enforces human-in-the-loop gates: conclusions and redlines always land with a human lawyer or manager for explicit sign-off. ### Solution: Autonomous reconciliation agents that match invoices to ledgers (slug: /solutions/finance) - **Category**: Finance & accounting - **Status**: Live (Runs Today) - **URL**: https://synthhires.com/solutions/finance (Spanish: https://synthhires.com/es/solutions/finance) - **Target Audience**: Finance controllers, accounting teams, and operations directors in high-transaction businesses. - **The Problem**: Month-end closing is a manual nightmare of spreadsheets, missing receipts, and multi-currency conversions. Traditional reconciliation software charges exorbitant per-invoice fees. - **Step-by-Step Autonomous Workflow**: 1. **Bank feed & invoice ingestion**: Statements (CSV, OFX) and vendor invoices arrive via email, SFTP, or direct upload. 2. **Deterministic matching**: Exact amount, date, and vendor ID matches are reconciled automatically in memory. 3. **Fuzzy resolution & exception gate**: Agent analyzes partial matches, currency fee deltas, and flags discrepancies for human review. 4. **Close package preparation**: Drafts balanced journal entries ready for your controller to sign off into your accounting system. - **Integrations**: Stripe Webhooks, Plaid API, Cloudflare Workers, PostgreSQL, Email Parser - **Market Incumbent**: Vic.ai (https://www.vic.ai) - Incumbent Pricing: Custom enterprise pricing starting at $1,000+/month with minimum invoice volume commits. - Where the Incumbent Wins: Native bi-directional enterprise ERP integrations (SAP, NetSuite, Coupa) with complex multi-subsidiary consolidation. - Where SynthHires Wins: BYOK wholesale model ($0-$9/mo platform fee vs seat licenses), zero lock-in, client-side encryption. - **Why Switch**: - Pure BYOK & credit consumption: pay fractions of a cent per reconciliation instead of $0.50-$2.00 vendor markups. - Open export: outputs standard journal entries for QuickBooks, Xero, Holded, or Excel. - **Honest Tradeoffs & When NOT to Use**: - Not a certified bank or tax filing agent: SynthHires automates triage and drafts, but a human accountant must file declarations. - **Verified FAQ**: - **Q: Can the agent connect directly to my bank?** A: Yes, via Open Banking / Plaid connectors configured with your own API credentials, or via standard CSV/OFX statement exports. - **Q: How does it handle ambiguous receipts?** A: Ambiguous items are routed to the approval gate with a proposed categorization. The agent never forces an unverified journal entry. - **Q: What formats can I export into?** A: Standard CSV, JSON, and ledger formats compatible with Xero, QuickBooks, Holded, and enterprise ERPs. ### Solution: IT service desk agents that triage alerts and diagnose logs (slug: /solutions/it) - **Category**: IT operations - **Status**: Live (Runs Today) - **URL**: https://synthhires.com/solutions/it (Spanish: https://synthhires.com/es/solutions/it) - **Target Audience**: IT directors, DevOps engineers, and system administrators managing growing company fleets. - **The Problem**: IT teams spend half their days answering the same password resets and VPN connection issues, while critical production alerts get lost in channel noise. - **Step-by-Step Autonomous Workflow**: 1. **Incident or ticket received**: Employee posts in #it-help, or Sentry/Datadog fires a webhook alert. 2. **Automated diagnostic execution**: Agent executes read-only MCP queries against cloud status, logs, and knowledge bases. 3. **Remediation script proposal**: Synthesizes root cause and drafts exact command or guided resolution steps. 4. **Admin approval & execution**: Engineer approves the action with a single click before any change is applied. - **Integrations**: Datadog, Sentry, Slack Webhooks, Cloudflare Workers, GitHub Actions - **Market Incumbent**: Moveworks (https://www.moveworks.com) - Incumbent Pricing: Annual enterprise contracts starting at $30,000+/year based on employee headcount. - Where the Incumbent Wins: Native deep connectors for enterprise Active Directory, ServiceNow, and Workday with pre-built enterprise device workflows. - Where SynthHires Wins: BYOK wholesale model ($0-$9/mo platform fee vs seat licenses), zero lock-in, client-side encryption. - **Why Switch**: - No per-employee seat tax: scale to thousands of users without paying $25-$50/employee/month. - Extensible via MCP: connect any internal CLI, Docker host, or internal API directly. - **Honest Tradeoffs & When NOT to Use**: - No native Windows Active Directory agent: complex on-prem domain controller scripts require your own custom MCP bridge. - **Verified FAQ**: - **Q: Can the agent execute terminal commands?** A: Only read-only diagnostics by default. Any mutating command requires an explicit Human-in-the-Loop gate with administrator approval. - **Q: Does it support Slack and Microsoft Teams?** A: Yes. Inbound triggers and approval notifications work natively across Slack and Teams webhooks. - **Q: Where do error logs get processed?** A: Logs are filtered in ephemeral Cloudflare Workers memory and sent to your configured model via BYOK with zero server storage. ### Solution: HR onboarding and employee handbook assistants with zero hallucination (slug: /solutions/hr) - **Category**: People operations - **Status**: Live (Runs Today) - **URL**: https://synthhires.com/solutions/hr (Spanish: https://synthhires.com/es/solutions/hr) - **Target Audience**: People operations leads, HR managers, and founders looking to give new hires immediate answers. - **The Problem**: HR managers answer the same 10 handbook questions 50 times a month, while employees wait days for simple policy confirmations or leave requests. - **Step-by-Step Autonomous Workflow**: 1. **Employee inquiry arrives**: Staff member asks in Slack, email, or internal portal (e.g. "What is our parental leave policy?"). 2. **Handbook RAG lookup**: Agent retrieves exact matching policy from your Notion or PDF handbook with page citations. 3. **Pre-filled request draft**: If an action is needed (vacation request, stipend), the agent drafts the submission. 4. **Manager approval gate**: Direct manager receives Slack button to approve or reject with one click. - **Integrations**: Slack, Notion MCP, Google Workspace, PostgreSQL, Email - **Market Incumbent**: Rippling (https://www.rippling.com) - Incumbent Pricing: Starts at $8/employee/month plus required base platform fees and per-module add-ons. - Where the Incumbent Wins: Integrated global payroll, health insurance administration, and corporate laptop provisioning in one suite. - Where SynthHires Wins: BYOK wholesale model ($0-$9/mo platform fee vs seat licenses), zero lock-in, client-side encryption. - **Why Switch**: - Zero per-employee seat tax: use your own knowledge base and API keys without paying monthly fees that scale with your headcount. - Direct sync with your current Notion or Google Docs wiki without forcing a payroll migration. - **Honest Tradeoffs & When NOT to Use**: - We do not run payroll or tax withholdings: we automate knowledge retrieval, workflows, and triage. - **Verified FAQ**: - **Q: Can we restrict sensitive policies to specific roles?** A: Yes. Knowledge collections can be restricted to specific Slack channels or user groups so executive compensation or termination policies are never public. - **Q: What formats can the employee handbook be in?** A: Notion databases, Google Docs, Markdown repositories, or uploaded PDF documents. - **Q: Does it support multi-language teams?** A: Yes. An employee can ask in Spanish, English, French, or German and receive accurate answers grounded in your English handbook. ### Solution: B2B lead enrichment and account research without credit markups (slug: /solutions/revenue) - **Category**: Revenue operations - **Status**: Live (Runs Today) - **URL**: https://synthhires.com/solutions/revenue (Spanish: https://synthhires.com/es/solutions/revenue) - **Target Audience**: B2B sales leaders, revenue operations managers, and agency founders managing outbound pipelines. - **The Problem**: Lead enrichment tools charge punitive credit prices for public information, while sales reps waste 40% of their day manually copying data from LinkedIn and company websites. - **Step-by-Step Autonomous Workflow**: 1. **New lead captured**: A demo request arrives via website form, CRM webhook, or uploaded lead list. 2. **Autonomous web research**: Agent queries Brave Search API, company landing pages, and tech stack signals. 3. **Account dossier compilation**: Synthesizes headcount, funding, key challenges, and personalized value propositions. 4. **Sales briefing & approval gate**: SDR receives one-page briefing in Slack or CRM with draft personalized outreach for review. - **Integrations**: HubSpot CRM, Brave Search API, PostgreSQL, Resend, Slack - **Market Incumbent**: Clay (https://www.clay.com) - Incumbent Pricing: Tiered monthly plans from $149/mo to $800+/mo based on credit consumption for third-party enrichment. - Where the Incumbent Wins: Over 50 pre-aggregated data vendor integrations with a spreadsheet-like waterfall enrichment user interface. - Where SynthHires Wins: BYOK wholesale model ($0-$9/mo platform fee vs seat licenses), zero lock-in, client-side encryption. - **Why Switch**: - No monthly credit limits: enrich 50 or 5,000 leads without sudden tier upgrade penalties. - Full prompt and logic transparency: tweak the research questions and data points anytime. - **Honest Tradeoffs & When NOT to Use**: - We do not sell proprietary B2B email databases: you bring the lead email/domain, and we research the company. - **Verified FAQ**: - **Q: Where does the company information come from?** A: From public company websites, official press releases, job boards, and real-time web search APIs using your own API credentials. - **Q: Can it write back to HubSpot or Salesforce?** A: Yes. Via standard webhook triggers or MCP connectors configured with your CRM tokens. - **Q: Does it send cold emails automatically?** A: By default, all generated outreach requires an approval gate where the account executive clicks to review and approve before sending. ## Comparative Benchmarks & Incumbent Analysis (Reviewed: 2026-09-15) SynthHires maintains verified, factual comparison matrices linked to primary source documentation: ### Comparison: SynthHires vs OpenCode (slug: /compare/opencode) - **Vendor**: OpenCode / Anomaly (https://opencode.ai/) - **URL**: https://synthhires.com/compare/opencode (Spanish: https://synthhires.com/es/compare/opencode) - **Headline**: SynthHires vs OpenCode: Agent Platform vs Terminal Coding Agent - **Overview**: An open-source AI coding agent that runs in your terminal, with desktop and IDE builds available. - **Last Verified Date**: 2026-09-17 - **Where OpenCode Wins**: - It is open source: you can read, audit, fork and self-build the agent you actually run. - A terminal agent fits an existing git-and-shell workflow with no platform account in between. - Provider configuration is a single config file, and the agent is model-agnostic by design. - **Where SynthHires Wins**: - Multi-agent orchestration is the core abstraction, not a single interactive session: parallel read-only batches and serialized writers are part of the engine. - Governance is a product surface: approval gates, policy enforcement and audit trails around agent operations for teams that need them. - Nothing to install to start: the whole workspace is in the browser, with auto-named and persisted conversations. - **Verified Primary Source Hosts**: opencode.ai - **Common Questions**: - **Q: Is OpenCode a SynthHires alternative?** A: Only for part of the job. Both let you bring your own model keys, and both speak MCP. OpenCode is a coding agent you drive from your terminal; SynthHires is a browser-based platform for running and supervising multiple agents with approval gates. If your work is editing a repository interactively, OpenCode is the more direct tool; if your work is a supervised pipeline of agents over a codebase, SynthHires targets that. - **Q: Does SynthHires have an open-source agent like OpenCode?** A: No — SynthHires is a proprietary platform. That is a real difference and it is stated in the comparison table rather than glossed over. What you do control is your keys: they are encrypted client-side with AES-256-GCM and never reach our servers. - **Q: Can I use OpenCode and SynthHires together?** A: Yes, and the two choices are not exclusive. OpenCode runs on your machine against your repository; SynthHires runs agents in the browser and can attach a local runtime through its Bridge daemon. Teams commonly keep a terminal agent for hands-on edits and use a supervised platform for the repeatable work. ### Comparison: SynthHires vs Cursor (slug: /compare/cursor) - **Vendor**: Anysphere (https://cursor.com/) - **URL**: https://synthhires.com/compare/cursor (Spanish: https://synthhires.com/es/compare/cursor) - **Headline**: SynthHires vs Cursor: BYOK Agent Platform vs AI Code Editor - **Overview**: A closed-source AI-first code editor built around deep integration with your local editor workflow, with cloud agents available from the Pro tier up. - **Last Verified Date**: 2026-09-15 - **Where Cursor Wins**: - Editor integration remains the deepest in the category: inline diffs, context-aware editing and codebase indexing in one surface. - Cloud agents and Bugbot run on the same subscription as the editor, so there is no second account to manage. - Privacy mode is a documented switch: code data is then not used for training by Cursor or its model providers. - **Where SynthHires Wins**: - No per-seat subscription for the agent layer: the free tier costs $0 and the paid tier is $9/month, while your tokens are billed by your own provider with no markup from us. - Keys are encrypted in the browser with AES-256-GCM and are never sent to SynthHires servers — the provider is called with the key from your device. - Agents are not confined to an editor session: they can be supervised as a workforce, with approval gates and audit over their operations. - **Verified Primary Source Hosts**: cursor.com - **Common Questions**: - **Q: Is SynthHires cheaper than Cursor?** A: It depends on what you are comparing. Cursor bundles editor features and a monthly model allowance into a subscription starting at $20/month per user. SynthHires charges $0 on the free tier and $9/month on the paid tier, and never marks up tokens: you pay your own provider at their published rate. The honest comparison is total cost for your workload, and that is why the table above lists both the subscription and the token handling separately. - **Q: Where do my API keys live with SynthHires?** A: In your browser, encrypted with AES-256-GCM. The keys are never sent to SynthHires servers; requests go from your device to the model provider. That is a different model from a cloud editor that keeps your credentials server-side. - **Q: Can I run agents outside the editor with SynthHires?** A: Yes. Agents run in the browser, can be given a repository workspace, and can attach a local runtime through the Bridge daemon. That covers the long-running, supervised work that does not belong in an interactive editing session. ### Comparison: SynthHires vs Pi (slug: /compare/pi) - **Vendor**: Earendil Works (https://pi.dev/) - **URL**: https://synthhires.com/compare/pi (Spanish: https://synthhires.com/es/compare/pi) - **Headline**: SynthHires vs Pi: Supervised Agent Platform vs Minimal Terminal Harness - **Overview**: A minimalist, MIT-licensed terminal coding harness distributed as a toolkit of packages (agent core, unified multi-provider LLM API, TUI, CLI). - **Last Verified Date**: 2026-09-15 - **Where Pi Wins**: - The MIT license plus a small core means you can vendor, modify or embed it without asking anyone. - The packages are usable standalone — the unified LLM API and the agent runtime are libraries, not just parts of a CLI. - Security posture is documented honestly: it ships without a permission system by default, and their docs give three concrete containerization patterns. - **Where SynthHires Wins**: - Our engine ships with guardrails rather than expecting you to containerize it: destructive-command guards, verification gates and approval checkpoints around tool execution. - Multi-agent orchestration and supervision are product features, not something you assemble from a runtime library. - Team-facing surfaces (audit, governance, persistent conversation history) come with the platform. - **Verified Primary Source Hosts**: pi.dev, github.com - **Common Questions**: - **Q: Is Pi safer than SynthHires because it is open source?** A: Auditability and isolation are different properties. Pi being MIT-licensed means you can read every line, which is real value. Its own documentation states it does not include a permission system for filesystem, process, network or credential access and runs with the permissions of the launching user, recommending containerization for stronger boundaries. SynthHires is not open source; it ships guardrails instead, including destructive-command guards and approval gates before sensitive tool calls. - **Q: Can I use Pi as a library and SynthHires as the platform?** A: That combination is normal. Pi packages such as the unified LLM API and the agent runtime are meant to be embedded. SynthHires is a platform you use rather than embed, so the decision is usually about whether you want to own the harness code or the supervised workspace. - **Q: Does SynthHires support the same providers as Pi?** A: Both are BYOK and both target the major providers. SynthHires reads its model catalog (context windows and pricing) from a maintained registry, and encrypted keys stay in your browser; Pi configures providers through its unified API on your machine. ### Comparison: SynthHires vs Hermes Agent (slug: /compare/hermes-agent) - **Vendor**: Nous Research (https://hermes-agent.nousresearch.com/) - **URL**: https://synthhires.com/compare/hermes-agent (Spanish: https://synthhires.com/es/compare/hermes-agent) - **Headline**: SynthHires vs Hermes Agent: Team Platform vs Self-Hosted Agent - **Overview**: An open-source, self-hosted AI agent with persistent memory, self-created skills and a messaging gateway for Telegram, Discord and Slack. - **Last Verified Date**: 2026-09-15 - **Where Hermes Agent Wins**: - A messaging gateway is a genuinely useful interaction model: you talk to your agent from Telegram, Discord or Slack instead of opening a web app. - Its learning loop is the differentiator — skills created from experience and refined during use, so the agent accumulates capability over time. - MIT-licensed and fully self-hosted: it runs on your hardware, with local inference through tools like Ollama or vLLM if you want no vendor in the loop. - **Where SynthHires Wins**: - Your keys stay in your browser: encrypted client-side with AES-256-GCM and never sent to our servers, so adopting a hosted platform does not mean handing over credentials. - Governance and audit are product surfaces, which matters when several people operate the same agents. - No installation and no host to maintain: the workspace, agent history and forkable conversations live in the browser, with a local runtime available through the Bridge daemon when you want it. - **Verified Primary Source Hosts**: hermes-agent.nousresearch.com, nousresearch.com - **Common Questions**: - **Q: Hermes Agent or SynthHires for a personal assistant?** A: If you want an agent you reach from Telegram or Slack, with memory and skills that grow on your own hardware, Hermes Agent is built for exactly that and SynthHires is not. SynthHires is a supervised workspace: browser-based agents, approval gates and audit, aimed at repeatable work rather than a personal assistant. - **Q: Do both keep my API keys under my control?** A: Both give you control, by different mechanisms. Self-hosting Hermes Agent means the keys sit on your infrastructure. With SynthHires the keys stay in your browser, encrypted with AES-256-GCM, and never reach our servers — so the platform being hosted does not change who holds the credential. - **Q: Is SynthHires open source like Hermes Agent?** A: No. Hermes Agent is MIT-licensed; SynthHires is a proprietary platform. The comparison table lists that difference explicitly instead of leaving a reader to discover it later. ### Comparison: SynthHires vs Roomote (slug: /compare/roomote) - **Vendor**: Roo Code (https://github.com/RooCodeInc/Roomote) - **URL**: https://synthhires.com/compare/roomote (Spanish: https://synthhires.com/es/compare/roomote) - **Headline**: SynthHires vs Roomote: Agent Workforce vs Cloud Coding Agent - **Overview**: A source-available cloud coding agent that takes a task from Slack, Teams, Telegram, Discord or a web UI, works in an ephemeral sandbox and opens a pull request. - **Last Verified Date**: 2026-09-15 - **Where Roomote Wins**: - The chat-to-pull-request loop is unusually complete: it clones the repo, makes changes, runs tests, takes a screenshot and opens the PR with a preview link. - Ephemeral sandboxes per task (Modal, E2B, Daytona, Blaxel or local Docker) keep agents off the host machine entirely. - Source-available and self-hostable, with a published free tier for up to 10 users and no per-seat SaaS pricing on self-hosted deployments. - **Where SynthHires Wins**: - The workforce is not limited to code: the same agents handle research, documents, media and connected-device work through MCP tools. - Keys are encrypted client-side in the browser rather than stored by the deployment. - It is a managed product: there is no Postgres, Redis and sandbox provider to stand up before your first task. - **Verified Primary Source Hosts**: github.com - **Common Questions**: - **Q: Roomote or SynthHires for automating pull requests?** A: For pure pull-request automation from a chat tool, Roomote is purpose-built and its flow — sandbox, tests, screenshot, PR — is more specialised than ours. SynthHires is broader: the same agent platform handles research, documents and connected-device work, and can be supervised with approval gates. - **Q: Which one is cheaper to run?** A: Self-hosted Roomote is free for up to 10 users but you supply the infrastructure: Postgres, Redis, a sandbox provider and a host. SynthHires has no infrastructure to run: $0 on the free tier, $9/month on the paid tier, and tokens billed by your own provider. Compare total cost of ownership, not the licence line alone. - **Q: Does SynthHires only work on code?** A: No. Code workspaces are one integration among several; agents also run research, document generation, media tasks and local desktop automation through MCP tools and the Bridge daemon. ### Comparison: SynthHires vs OpenAI Codex (slug: /compare/codex) - **Vendor**: OpenAI (https://openai.com/codex/) - **URL**: https://synthhires.com/compare/codex (Spanish: https://synthhires.com/es/compare/codex) - **Headline**: SynthHires vs OpenAI Codex: BYOK Platform vs OpenAI Agent - **Overview**: OpenAI’s own coding agent, shipped as an Apache-2.0 licensed CLI for your terminal plus an IDE extension, a desktop app and a hosted agent inside ChatGPT. - **Last Verified Date**: 2026-09-15 - **Where OpenAI Codex Wins**: - The CLI repository is Apache-2.0, so the agent you actually run on your machine can be read, built and audited — and it installs from npm, Homebrew or a signed binary. - It is one product across four surfaces — terminal, editor, desktop app and ChatGPT — all connected to the same account, so a task can start in one and be reviewed in another. - Scheduled, always-on work is a first-class feature: their product page describes scheduling Codex for issue triage, alert monitoring and CI/CD. - **Where SynthHires Wins**: - Model choice is your catalog, not one vendor’s lineup: SynthHires is BYOK across the providers we list, while Codex runs on OpenAI’s models. - You do not hand a model provider account to a platform: your key is encrypted in your browser and requests go from your device to your provider. - Agents here are not only code agents: the same workspace runs research, documents and connected-device work under the same approval and audit surfaces. - **Verified Primary Source Hosts**: github.com, openai.com - **Common Questions**: - **Q: Is OpenAI Codex a SynthHires alternative?** A: Partly. Both run agents over a codebase, and both have a terminal path. Codex is OpenAI’s own agent running on OpenAI’s models across the terminal, editor, desktop and ChatGPT; SynthHires is a browser platform that is model-agnostic — you bring a key from whichever provider you prefer — and supervises multiple agents with approval gates. If you are all-in on OpenAI models, Codex is the more integrated choice. - **Q: Can I use my own model provider with Codex?** A: Codex is OpenAI’s agent on OpenAI’s models. Their README describes two ways in: signing in with ChatGPT, or configuring an OpenAI API key (which it calls additional setup). Choosing among many providers with your own keys is the premise SynthHires is built on instead. - **Q: Is the Codex CLI really open source?** A: Their repository states it is licensed under Apache-2.0 and lists npm, Homebrew and signed-binary installs, so the CLI that runs on your machine is open source. Codex inside ChatGPT is a separate hosted product and is not covered by that licence — the two are often conflated. ### Comparison: SynthHires vs Claude Code (slug: /compare/claude-code) - **Vendor**: Anthropic (https://code.claude.com/docs/en/overview) - **URL**: https://synthhires.com/compare/claude-code (Spanish: https://synthhires.com/es/compare/claude-code) - **Headline**: SynthHires vs Claude Code: Agent Platform vs Anthropic CLI - **Overview**: Anthropic’s agentic coding tool: it reads your codebase, edits files and runs commands, and it runs in your terminal, an IDE extension, a desktop app or the browser. - **Last Verified Date**: 2026-09-15 - **Where Claude Code Wins**: - Four surfaces under one subscription: terminal CLI, VS Code and JetBrains extensions, a desktop app, and a browser version that needs no local setup. - Subagents are a documented, first-class feature: each runs in its own context window with its own system prompt, tool access and permissions, and their docs also cover agent teams — sessions Claude spawns and supervises. - The operational detail is written down: native, Homebrew, WinGet and apt/dnf/apk installs, plus an honest note that only the native installs update themselves in the background. - **Where SynthHires Wins**: - Model choice is not tied to one vendor: SynthHires is BYOK across the whole catalog, while Claude Code’s own docs note that most of its surfaces require a Claude subscription or an Anthropic Console account. - Keys stay in the browser, encrypted client-side with AES-256-GCM and never sent to our servers, rather than being held by the platform you are signed into. - Governance is a product surface rather than a settings file: approval gates, policy enforcement and audit trails over agent operations for teams that need oversight. - **Verified Primary Source Hosts**: code.claude.com, claude.com - **Common Questions**: - **Q: Is Claude Code a SynthHires alternative?** A: For a single developer editing a repository, Claude Code is the more direct tool and its subagent model is genuinely well documented. SynthHires targets a different job: a browser workspace where several agents run over a codebase under approval gates and audit trails, with any provider’s key rather than one vendor’s subscription. - **Q: Do I need a Claude subscription to use Claude Code?** A: Their docs say most surfaces require a Claude subscription or an Anthropic Console account, while the terminal CLI, VS Code and JetBrains also support third-party providers. Claude Code is included in the Pro plan ($17/month billed annually, $20 monthly, per claude.com/pricing) and in Team standard seats. SynthHires has no vendor subscription at all: $0 on the free tier, $9/month on the paid tier, with your own provider key and no markup on tokens. - **Q: Can both run multiple agents?** A: Yes, with different shapes. Claude Code documents subagents with their own context windows and permissions, plus agent teams of supervised sessions. SynthHires runs parallel read-only batches with serialized writers, so concurrent work cannot collide on the same file, and puts approval gates on the operations that need a human. ### Comparison: SynthHires vs GitHub Copilot (slug: /compare/github-copilot) - **Vendor**: GitHub (https://github.com/features/copilot/plans) - **URL**: https://synthhires.com/compare/github-copilot (Spanish: https://synthhires.com/es/compare/github-copilot) - **Headline**: SynthHires vs GitHub Copilot: Platform vs Editor Assistant - **Overview**: GitHub’s AI coding assistant built into editors, the CLI and github.com, with a cloud agent that can take an issue, do the work and open a pull request. - **Last Verified Date**: 2026-09-15 - **Where GitHub Copilot Wins**: - Distribution is hard to beat: one plan covers VS Code, Visual Studio, JetBrains, Xcode, Neovim, Eclipse, Zed, SQL Server Management Studio, Raycast, the CLI and github.com. - There is a real free tier with no card required: 2,000 completions a month, the CLI, model choice and community support. - The cloud agent and pull-request code review are included from the Pro tier, and Pro+ can delegate tasks to third-party coding agents such as Claude Code and Codex in preview. - **Where SynthHires Wins**: - No credits to meter: you pay your own provider at their published rate with no markup, instead of tracking an included monthly credit balance. - Your key never leaves your browser — encrypted with AES-256-GCM, with requests going from your device to the provider — whereas Copilot is a hosted service where you supply no key at all. - Agent work is the surface, not a feature inside an editor: parallel read-only batches, serialized writers, approval gates and audit trails are the platform. - **Verified Primary Source Hosts**: github.com - **Common Questions**: - **Q: Is GitHub Copilot an alternative to SynthHires?** A: They overlap in the middle and differ at the edges. Copilot is an assistant and a cloud agent inside GitHub’s ecosystem: unbeatable if your work is already there. SynthHires runs a supervised team of agents from the browser over code and non-code work, with your own model keys. Choosing Copilot is choosing convenience and one bill; choosing SynthHires is choosing provider choice and visibility of the real token cost. - **Q: Do I need my own API key with GitHub Copilot?** A: No. Copilot is a subscription service with included GitHub AI Credits, and you can buy more on Pro and above — there is no key to manage. SynthHires goes the other way: bring a key from any provider in our catalog, no markup on tokens, $0 free tier and $9/month paid. If you want one bill and no key management, that is Copilot’s model. - **Q: Which one can run agents unattended?** A: Copilot’s cloud agent is purpose-built for that: assign it an issue and it researches, plans, writes the code and can open a pull request, with code review available too. SynthHires is built for the supervised middle ground — long-running pipelines with approval gates on the operations you want a human to see, rather than fully unattended merges. ## Documentation Index Core technical documentation for developers, engineers, and platform operators: - **Overview** (`/docs/overview`): What the Space is, how the local-first architecture works, and the core concepts behind every page. (Group: start) - **AI concepts from scratch** (`/docs/ai-concepts`): A plain-language primer on the ideas behind every page: tokens, context windows, sampling, embeddings, RAG, agents, tools, MCP and streaming. (Group: start) - **Architecture** (`/docs/architecture`): Full stack breakdown: Astro + TanStack Router + React, Cloudflare Workers, Drizzle + Postgres and the AI SDK v4 pipeline. (Group: start) - **Chat & orchestration** (`/docs/chat`): The conversation surface, streaming, agent orchestration, forks and the local-first persistence model. (Group: workspace) - **Agents, skills & prompts** (`/docs/agents`): The 16-agent catalog, agent hierarchies, reusable skills and the prompt library with slash triggers. (Group: workspace) - **Memory & knowledge** (`/docs/memory`): Cross-session recall, the semantic knowledge base (RAG), document ingestion and artifacts. (Group: workspace) - **Pipeline studio** (`/docs/studio`): Visual workflow orchestration: nodes, edges, plan extraction and workflow runs persisted in Postgres. (Group: workspace) - **Integrations & MCP** (`/docs/integrations`): Core connectors (GitHub, GitLab, Cloudflare, Slack), the MCP server registry, plugins and inbound channels. (Group: connectivity) - **Runtimes & bridge** (`/docs/runtimes`): The desktop/mobile bridge daemon, the wire protocol, device pairing, cloud sandboxes and Docker/SSH. (Group: connectivity) - **Get API keys** (`/docs/api-keys`): Step-by-step, up-to-date guide to getting an API key for every supported provider — OpenAI, Anthropic, Gemini, Groq, xAI, DeepSeek, Mistral, OpenRouter and more — plus web services like Firecrawl. (Group: connectivity) - **Models & API keys** (`/docs/models`): The models.dev-synced catalog, every supported provider, the BYOK AES-256-GCM vault and global inference parameters. (Group: connectivity) - **Content studio** (`/docs/content`): Generative image, video, audio and 3D pipelines, plus the searchable content library. (Group: content) - **Usage, health & teams** (`/docs/governance`): Cost and token metrics, P95 latency and success rates, cron jobs, notifications and collaborative teams. (Group: platform) - **Security & BYOK** (`/docs/security`): The BYOK promise, local AES-256-GCM encryption, single-use handoff envelopes, auth and request hardening. (Group: platform) ## Entity Graph & Verification Anchors - **Legal Entity**: SynthHires, operated by PlanoBinario - **Parent Organization**: PlanoBinario (https://planobinario.com) - **Primary Website**: https://synthhires.com - **Public Inquiries & Support**: synthhires@planobinario.com - **Social & Code Profiles**: - X: https://x.com/SynthHires - GitHub: https://github.com/planobinario - LinkedIn: https://linkedin.com/company/planobinario - **Languages**: Full parity in English (en) and Spanish (es).