Stop guessing
timelines.
Get data-driven estimates your AI can actually use.
26 MCP tools. 30-second install. Works with Claude Code, Cursor, Codex CLI, Cline, Zed, and any MCP client.
Sound Familiar?
You've tried everything else
Your AI is powerful. But when it comes to "how long will this take?" — it guesses. Badly.
Asking AI directly
"This will take 2 hours." Two days later, you're still waiting. LLMs hallucinate timelines because they can't perceive time.
Prompt engineering
You craft the perfect estimation prompt. It works once. The next time, it gives a completely different answer. Inconsistent and unreliable.
Gut feelings
Software projects run 2–10x over estimate. LLMs replicate and amplify human optimism bias. The result: agentic overconfidence.
Research backs this up: A UPenn study showed 4% deal closure under wall-clock limits vs 99% under turn-based limits. Injecting explicit time state produced an 8x improvement.
The Solution
40 years of estimation research,
packaged for AI
We studied PERT, COCOMO II, Monte Carlo, reference class forecasting — every major estimation method — and distilled them into 26 structured tools that AI agents can call like functions.
- AI estimates in token budgets, not minutes
- Events 10 tokens apart are identical whether 1 second or 1 hour apart
- No feedback loop — the same errors repeat forever
- Grounded estimates using proven algorithms, not guesses
- 30-second install, zero config, no API keys
- Self-improving — gets more accurate with every estimate
The Value Equation
Dream Outcome
Estimates your team trusts
Perceived Likelihood
COCOMO II, PERT, Monte Carlo — proven methods
Time Delay
30-second install, instant results
Effort
AI calls the tools automatically, zero config
Live Demo
See it in action
A real interaction. No cherry-picking, no edits.
How long will a 15K LOC project take with a team of 5?
Based on COCOMO II estimation:
Confidence: 72% · Method: COCOMO II + PERT blend
What You Get
26 tools. 6 layers. One install.
From basic time operations to Monte Carlo simulation. Each layer builds on the last.
The Guarantee
Free forever. Open source. No lock-in.
The code is public. Audit it yourself. Telemetry is off by default. Every data point is documented.
Off by Default
Telemetry requires explicit opt-in. Preview exactly what will be shared before enabling.
Zero PII
No names, emails, project names, or company data. Dates truncated to day-only.
Self-Hostable
Configurable endpoint. Run your own telemetry server. Full API contract documented.
Integration
Three surfaces, one API
All 26 tools are accessible through MCP, CLI, and REST API. Use whichever fits your workflow.
MCP Server
Stdio transport for Claude Code, Cursor, Codex CLI, Cline, Zed
CLI
Scriptable subcommands with JSON or table output
REST API
HTTP server with OpenAPI spec and AI discoverability
Full Reference
All 26 tools, one API
Zod-validated schemas, .describe() on every field, readOnlyHint annotations.
| Tool | Description |
|---|---|
| get_current_time | Current time in any IANA timezone |
| convert_timezone | Convert ISO-8601 between timezones |
| parse_duration | Parse "2h30m" into structured seconds |
| time_math | 6 operations: add_days, diff, format_duration, add_hours, add_minutes, add_weeks |
| add_business_days | Add N working days, skipping weekends and holidays (5 countries) |
| count_business_days | Count working days between two dates |
| pert_estimate | Three-point PERT Beta distribution: E = (O+4M+P)/6 |
| cocomo_estimate | COCOMO II with 5 LLM-adapted cost drivers |
| sprint_forecast | Sprint velocity forecasting with variance-based pessimistic estimate |
| critical_path | CPM with forward/backward pass and merge bias |
| monte_carlo_schedule | Monte Carlo simulation with seeded PRNG, P10-P95 percentiles |
| reference_class_estimate | Reference class forecasting with industry correction factors |
| calibrate_estimates | Team-specific correction factor calibration from historical data |
| token_time_bridge | Map token budgets to wall-clock time for 16 LLM models |
| token_cost_estimate | Token cost estimation with input/output breakdown |
| compare_models | Side-by-side comparison across 16 LLM models |
| accuracy_trend | Sliding-window MAPE tracking over time |
| schedule_risk | Confidence intervals using team MdAPE |
| record_actual | Submit actual hours for a previous estimate |
| feedback_health | Comprehensive feedback data quality report |
Get Started
Up and running in 30 seconds
No config files. No API keys. One command.
Install Epoch
Ask any time question
"What's the PERT estimate for a 3-month sprint?"
"How many business days until March 15?"
Get grounded, data-driven estimates
{
effort: "45 person-months",
duration: "8.3 months",
risk: "medium",
confidence: "72%"
}
30 seconds. No API keys. No config.
Install Epoch — It's FreeFAQ
Common questions
Built With