{"id":51090315,"url":"https://github.com/farjad-hasan/aidem-framework","last_synced_at":"2026-06-24T01:03:50.733Z","repository":{"id":358970373,"uuid":"1242180210","full_name":"farjad-hasan/aidem-framework","owner":"farjad-hasan","description":"AIDEM is an open framework for estimating, grooming, and calibrating AI-assisted engineering work.","archived":false,"fork":false,"pushed_at":"2026-05-19T20:09:32.000Z","size":28,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-05-19T23:49:11.094Z","etag":null,"topics":["agile","ai-assisted-development","ai-engineering","ai-governance","delivery-planning","engineering-leadership","engineering-management","jira","software-delivery","software-estimation"],"latest_commit_sha":null,"homepage":"","language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"other","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/farjad-hasan.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":"governance/remediation-playbooks.md","roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-05-18T07:38:14.000Z","updated_at":"2026-05-19T20:09:55.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/farjad-hasan/aidem-framework","commit_stats":null,"previous_names":["farjad-hasan/aidem-framework"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/farjad-hasan/aidem-framework","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/farjad-hasan%2Faidem-framework","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/farjad-hasan%2Faidem-framework/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/farjad-hasan%2Faidem-framework/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/farjad-hasan%2Faidem-framework/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/farjad-hasan","download_url":"https://codeload.github.com/farjad-hasan/aidem-framework/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/farjad-hasan%2Faidem-framework/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":34712581,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-26T15:22:16.424Z","status":"online","status_checked_at":"2026-06-23T02:00:07.161Z","response_time":65,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["agile","ai-assisted-development","ai-engineering","ai-governance","delivery-planning","engineering-leadership","engineering-management","jira","software-delivery","software-estimation"],"created_at":"2026-06-24T01:03:50.035Z","updated_at":"2026-06-24T01:03:50.723Z","avatar_url":"https://github.com/farjad-hasan.png","language":null,"funding_links":[],"categories":[],"sub_categories":[],"readme":"# AIDEM — AI-Assisted Delivery Estimation Model\n\nAIDEM is an open framework for estimating, grooming, and calibrating AI-assisted engineering work.\n\nIt helps teams avoid two common mistakes:\n\n1. Treating AI as a blanket productivity discount.\n2. Ignoring the review, testing, rework, flow, and risk costs that appear when AI enters the delivery path.\n\nAIDEM is not a precision formula. It is a **grooming and calibration protocol** for engineering teams working with AI.\n\n\u003e AI changes engineering work. AIDEM keeps planning honest.\n\n---\n\n## Current version\n\n**AIDEM v1.3 + Pilot Addendum + Dashboard Patch**\n\nThis version includes:\n\n- AIDEM-T for teams transitioning into AI-assisted delivery\n- AIDEM-N for AI-native teams using reference-class forecasting\n- AIDEM-H for mixed/hybrid work\n- Risk scoring with severity floor rules\n- Confidence calibration and range-width tracking\n- Lite vs Full grooming paths\n- Model/tool versioning and drift rules\n- Skill-development modes\n- Remediation playbooks for flagged pilot metrics\n\n---\n\n## Why AIDEM exists\n\nAI-assisted engineering has made traditional estimation harder, not easier.\n\nSome work becomes faster:\n\n- scaffolding\n- boilerplate\n- documentation\n- test drafts\n- refactoring with known patterns\n- code search and explanation\n\nBut some work does not disappear:\n\n- business judgment\n- requirements ambiguity\n- domain validation\n- code review\n- QA and regression testing\n- integration and deployment flow\n- production and data risk\n\nAIDEM gives teams a structured way to discuss those tradeoffs before committing to delivery expectations.\n\n---\n\n## Quick start (first 2 weeks)\n\nIf you are adopting AIDEM for the first time, use this order:\n\n1. Read `docs/aidem-framework.md` for mode definitions and risk logic.\n2. Run one sprint using `templates/lite-template.md` for low-risk tickets.\n3. Require `templates/full-template.md` for High/Critical or unclear work.\n4. At sprint end, review calibration and range width using `docs/dashboard-patch.md`.\n5. Apply remediations from `governance/remediation-playbooks.md` for any flagged metrics.\n\nRecommended pilot sequence is documented in `docs/how-to-pilot.md`.\n\n---\n\n## The three AIDEM modes\n\n### AIDEM-T — Transition Mode\n\nUse when the team has credible historical experience estimating similar work before AI assistance.\n\nFormula:\n\n```text\nT_planning = [\n  (T_AI_eligible × (1 - L))\n  + T_nonAI\n  + T_context\n  + T_review\n  + T_test\n  + T_rework\n  + T_flow\n] × R\n```\n\nWhere:\n\n- `T_AI_eligible` = work AI can realistically accelerate\n- `L` = AI leverage assumption\n- `T_nonAI` = work AI does not meaningfully reduce\n- `T_context` = time to provide context, examples, constraints, and repo understanding\n- `T_review` = human review of AI-assisted output\n- `T_test` = QA, regression, sandbox, or acceptance testing\n- `T_rework` = fixing wrong assumptions, hallucinated logic, or missed edge cases\n- `T_flow` = PR wait, handoffs, dependency coordination, and release overhead\n- `R` = risk multiplier\n\n### AIDEM-N — Native Mode\n\nUse when AI is ambient in the team’s workflow and there is no credible “without AI” baseline.\n\nFormula:\n\n```text\nT_planning = T_reference + T_validation + T_flow + T_risk\n```\n\nWhere:\n\n- `T_reference` = observed delivery range from similar completed AI-assisted work\n- `T_validation` = review, QA, security, domain, or compliance validation\n- `T_flow` = review wait, handoffs, release, and stakeholder flow\n- `T_risk` = uncertainty buffer based on production, customer, reversibility, or security risk\n\n### AIDEM-H — Hybrid Mode\n\nUse when a ticket contains both transition-style and AI-native sub-work.\n\nHybrid mode must name the split explicitly:\n\n```text\nTransition-mode sub-tasks:\n- ...\n\nNative-mode sub-tasks:\n- ...\n```\n\nNo duplicate overhead rule:\n\n\u003e Validation, flow, and risk are counted once at the whole-ticket level unless a sub-task has a distinct release path.\n\nIf AIDEM-H exceeds 30–40% of tickets in a pilot, review whether teams are overusing Hybrid to avoid clearer mode selection.\n\n---\n\n## AIDEM Lite vs AIDEM Full\n\n### AIDEM Lite\n\nUse for low-risk, familiar, reversible work.\n\nMinimum fields:\n\n- Mode: T / N / H\n- Estimate range\n- Confidence level\n- 30-second risk pre-check\n- Major blocker, if any\n\n### AIDEM Full\n\nRequired for:\n\n- High or Critical risk work\n- production data impact\n- customer-facing changes\n- irreversible changes\n- security/privacy concern\n- unclear architecture or discovery-heavy work\n- hybrid work with unclear boundaries\n\nFull mode includes the complete risk scorecard, confidence rationale, governance gates, model/tool metadata, and completion review.\n\n---\n\n## 30-second Lite/Full risk pre-check\n\nBefore choosing Lite, score the four 2x risk factors:\n\n| Factor | Score |\n|---|---:|\n| Production data impact | 0–3 |\n| Customer-facing impact | 0–3 |\n| Reversibility concern | 0–3 |\n| Security/privacy concern | 0–3 |\n\nRules:\n\n- Any factor scored **2 or higher** requires **AIDEM Full**.\n- Any factor scored **3** forces minimum **High** risk.\n- If all factors are 0–1, AIDEM Lite may be used if the work is familiar and reversible.\n\n---\n\n## Confidence and range width\n\nAIDEM tracks both calibration and sharpness.\n\n```text\nRange Width % = (High Estimate - Low Estimate) / Midpoint Estimate\n```\n\n| Confidence tier | Target max range width |\n|---|---:|\n| High | ≤ 25% |\n| Medium | ≤ 40% |\n| Low | ≤ 60% |\n\nEvery dashboard metric should also show sample size `n`.\n\nRule:\n\n\u003e Do not treat a tier-level calibration result as stable until it has at least 10 completed tickets.\n\n---\n\n## Repository structure\n\n```text\n/docs\n  aidem-framework.md\n  pilot-addendum.md\n  dashboard-patch.md\n/templates\n  jira-template.md\n  lite-template.md\n  full-template.md\n/examples\n  example-library.md\n/governance\n  risk-scorecard.md\n  remediation-playbooks.md\n```\n\n---\n\n## Suggested adoption checklist\n\n- [ ] Define when teams should default to AIDEM-T vs AIDEM-N.\n- [ ] Agree the team-level definition of \"actual effort\" before Sprint 1.\n- [ ] Decide who owns dashboard updates and weekly calibration review.\n- [ ] Set a minimum sample-size rule (`n \u003e= 10`) before policy changes.\n- [ ] Confirm escalation path for High/Critical risk tickets.\n\n---\n\n## Recommended pilot\n\nRun AIDEM for one quarter before making major framework changes.\n\nTrack:\n\n- estimate range vs actual\n- confidence calibration by tier\n- median range width by tier\n- sample size `n`\n- grooming time per ticket\n- AIDEM-H share of tickets\n- Lite vs Full usage\n- review/rework effort\n- defects and escaped defects\n- AI tooling incidents\n\n---\n\n## What AIDEM does not do\n\nAIDEM is downstream of portfolio selection.\n\nIt helps estimate, groom, govern, and calibrate work that has already entered the delivery pipeline. It does not decide whether AI has changed what work is worth doing. Portfolio selection, opportunity discovery, and product strategy require separate decision frameworks.\n\n---\n\n## License\n\nThis project is shared under the **Creative Commons Attribution 4.0 International License** for documentation and framework content.\n\nTemplates may be adapted freely with attribution.\n\n---\n\n## Contributing\n\nFeedback is welcome, especially from engineering leaders, tech leads, product leaders, and engineers using AI-assisted delivery in real teams.\n\nUseful contributions include:\n\n- pilot results\n- alternative risk scoring methods\n- examples from real engineering workflows\n- improvements to Lite vs Full usage\n- confidence calibration lessons\n- critiques of where the framework becomes too heavy\n\nSee [CONTRIBUTING.md](CONTRIBUTING.md) for contribution guidance.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Ffarjad-hasan%2Faidem-framework","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Ffarjad-hasan%2Faidem-framework","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Ffarjad-hasan%2Faidem-framework/lists"}