Home/Why Claude

Why we picked Claude.
Not GPT. Not open source. By choice.

We build agentic AI for production. That means agents that loop, call tools, handle exceptions, and ship work that customers pay for. After running our internal evals across the major frontier model families, we settled on Claude — Sonnet 4.6 for reasoning, Haiku 4.5 for routing, Opus 4.7 for the hard stuff. Here's the technical case, written for engineers who'll have to defend the choice internally.

01 · TOOL USE

Claude returns well-formed tool calls — every time.

Production agents loop. Loop means tool_use → execute → tool_result → tool_use again. If 1 in 50 tool calls is malformed, you spend the next month writing parsers and retry logic instead of shipping features. Claude Sonnet 4.6 is the most reliable model we've tested at parallel tool calls + structured output across long sessions.

In production: Our recruiter agent makes ~12 tool calls per candidate (search → enrich → score → draft email → log CRM). In our internal eval set, Claude Sonnet returned a malformed tool call at well under 0.1%, handled with a single retry. Earlier-generation models in the same eval hit roughly 6× that rate.

02 · MCP

MCP is the agent-tool standard. Anthropic invented it.

Model Context Protocol is the open standard that lets any AI client call any tool server. Cursor, Zed, Continue, our backend — they all speak MCP. That means a customer with an internal MCP server for their CRM can plug it into our agents without custom integration work. The market is converging on MCP. Building on it is a no-regret move.

In production: Our backend exposes an MCPConnector that wraps any MCP server (JSON-RPC 2.0). Today: 12 first-party MCP integrations (Gmail, QuickBooks, Slack, etc). Tomorrow: customer-registered MCP servers become first-class agent capabilities.

03 · PROMPT CACHING

90% cost reduction on the parts that don't change.

Every agent run starts with the same ~3K-token system prompt, tool definitions, RAG context, and few-shot examples. Without caching, you pay full input rate on all of it, every run. With Claude's prompt caching, the cached portion costs 10% of normal — which is what gets a run cost down to $0.003. No other frontier model offers caching at this granularity.

In production: On the AP bookkeeper agent, prompt caching saves us roughly $0.01 per run at production volumes. That margin is what makes $0.50-per-invoice outcome pricing work without taking a loss on the long tail.

04 · SAFETY

Constitutional AI matters when buyers are CFOs and CIOs.

Regulated buyers — finance, healthcare, EU enterprises — read your model card. Claude's constitutional approach, low rate of confabulation on structured business tasks, and conservative default behavior on ambiguous instructions are reasons we close deals our competitors don't. It's not just a checkbox — it's why our agents refuse to send the email when the input is suspicious.

In production: On invoice processing, Claude flags ambiguous vendor-name mismatches as 'needs human review' significantly more often than earlier-generation comparison models in our internal eval. Fewer silent miscategorizations. Fewer post-deploy fires.

05 · EXTENDED THINKING

Sonnet 4.6 reasons before it acts. That changes the failure mode.

Extended thinking lets the model produce internal reasoning before its final answer. For agents, that means fewer 'confidently wrong' tool calls. We use it on the bookkeeping agent before posting to QuickBooks, on the recruiter agent before sending an outreach email, and on the voice agent before classifying a call outcome.

In production: Adding extended thinking on AP categorization measurably cut our human-override rate in internal evaluation — meaningfully shrinking the approval queue per customer per day. We publish a current methodology + numbers under NDA.

06 · OPUS WHEN YOU NEED IT

Three model sizes, one API, one mental model.

Haiku 4.5 for routing ($1/M in, $5/M out). Sonnet 4.6 for reasoning ($3/M, $15/M). Opus 4.7 for the hardest tasks ($15/M, $75/M). Same API, same tool format, same safety properties. We route between them inside a single agent run — cheap classification followed by expensive reasoning followed by cheap formatting — without any infrastructure switch.

In production: Our recruiter agent uses Haiku to filter 500 candidates down to 20, Sonnet to rank and personalize outreach for those 20, and Opus only on the 2-3 'need a really thoughtful intro email' candidates. Effective per-candidate cost lands in the low single-digit cents at production volume.

The objections we hear weekly

But GPT-4 is the most popular — won't I have more flexibility?

Popularity is not durability. The agentic AI market is converging on MCP, on prompt caching, on structured tool use — all of which Claude leads on. Picking the popular model today often means rewriting in 12 months when the platform shifts.

What if Anthropic raises prices?

Anthropic has cut prices on every model release since Claude 3 (March 2024). Sonnet 4.6 is half the price of Claude 3 Opus at higher quality. We pass through 100% of cost changes, up or down, in our outcome pricing — your $/result stays in your control regardless.

What about latency?

Haiku 4.5 first-token latency is ~280ms — below what voice agents need. Sonnet 4.6 streaming throughput is ~80 tokens/sec. Both are fast enough for real-time human-in-the-loop UX. We use Haiku for anything user-facing and Sonnet for reasoning-bound async work.

Can you swap in OpenAI if a customer demands it?

No — and that's a feature, not a bug. We optimize hard for Claude. A 'multi-model' platform is one that does each model poorly. If you need OpenAI specifically, we'll happily refer you elsewhere.

“Why not just use the Claude API directly?”

You could. So could anyone with an API key. To get to production you'd need:

  • — A run console showing every tool call, cost, and outcome, scoped per tenant
  • — Per-tenant budget caps that actually halt runs (not just alert)
  • — A hash-chained audit log that survives a regulator's “prove this didn't change” question
  • — An outcome ledger Stripe-meter-events synced for billing
  • — Twilio voice + WhatsApp + Slack human-in-the-loop with reply-1-to-send UX
  • — Gmail OAuth, Stripe webhooks, QuickBooks OAuth, Greenhouse webhooks, LinkedIn API plumbing
  • — RAG pipelines with vendor-history lookups, fallback, cache, version pinning
  • — Self-healing: monitor logs, diagnose failures, generate fixes, alert with what-broke / what-we-fixed

That's four engineers for six months before your first production agent. We did it. We run it for you. You ship agents on day five.

Book the audit →