A structured close-out that captures what actually happened: where the estimate was wrong, which risks bit, what the client valued most, and the two changes that would make the next engagement of this type better. Describe the finished project and a Taskade Genesis app runs the retrospective and files the lessons where they will be found again.
The Internal Docs App is a live Taskade Genesis app embedded on this page. It is a structured document store rather than a retrospective tool, so use it as the starting point: try it here, then click "Use this app" to clone it in about ten seconds and describe your own engagement so each document becomes a close-out retrospective with lessons and owners.
Consulting improves through repetition only if the repetition is examined. Most firms skip the retrospective because the next project has already started, which is precisely why the same estimating mistake happens four times. Making it a fifteen-minute structured routine is the whole trick. Workspace DNA connects it to the future. Memory holds the estimate, the actuals, the risk register, and the client feedback in one place. Intelligence, backed by 15+ frontier models from OpenAI, Anthropic, Google, and open-weight providers, compares planned against actual, names the pattern across engagements, and drafts the lesson in a form your next proposal can use. Execution files the lesson against the service line so it appears the next time you scope similar work.
The retrospective renders in whichever layout you need:
- List view for the guided close-out you work through with the team
- Table view for the estimate against actual comparison by workstream
- Board view for lessons moving from raised to agreed to applied
- Mind Map view for seeing recurring themes across engagements
- Plus the rest of the 7 project views
Automations make it happen at all. Connect Taskade automations so a final invoice paid opens the retrospective with the delivery data already loaded, sends the client feedback request, and books fifteen minutes in your calendar to complete it. With 100+ bidirectional integrations, forms and Gmail pull client feedback in while Slack pushes the reminder and the summary out.
The AI agents inside the app carry 34 built-in tools including persistent memory, file analysis, web search, custom slash commands, and multi-agent collaboration. Ask the agent which of your last ten engagements shared a root cause and you get a genuinely uncomfortable and genuinely useful answer.
Clone it, run it every time, and the learning is yours. Over a year it becomes the most valuable pricing and scoping data your firm owns.
Build your own close-out in Taskade Genesis, see live review apps in the Community Gallery, or learn scheduled runs in Learn. Compare against Scope and Price a New Engagement, and feed the lessons into Build a Reusable Consulting Asset Library.
Frequently Asked Questions
Should the client be part of the retrospective?
Run two: a short client-facing review that asks what worked and what they would change, and an internal one that is candid about estimating, staffing, and scope. Mixing them produces a polite conversation that teaches you nothing.
What questions matter most?
Where the estimate was wrong and why, which risks materialized, what the client valued that you did not expect, what you delivered that nobody used, and what you would price differently. Those five cover most of the value and take fifteen minutes.
How do I get honest feedback from a client?
Ask specific questions rather than general ones, and ask about the process rather than about you. "Which part of the engagement felt slowest" gets a real answer where "were you satisfied" gets a polite nine out of ten.
How do lessons actually get applied?
By being attached to a service line rather than filed in a document. When you scope the next engagement of that type, the relevant lessons surface in the scoping app, which is the only mechanism that reliably beats good intentions.
What if the engagement went badly?
That is the one worth the most. Record it factually, separate what you controlled from what you did not, and turn each controllable factor into a change to scoping, staffing, or qualification. The alternative is repeating it with a different client.
Can this feed my proposals?
Directly. Real durations, real effort, and real client-valued outcomes make the next proposal both more accurate and more persuasive, because you can describe what typically happens rather than what you hope will happen.
How long should a retrospective take?
Fifteen minutes for a small engagement with the data pre-loaded, an hour for a large one with several people. The automation matters more than the format, because the retrospective that gets scheduled is the one that gets done.
What should I do with a lesson once it is recorded?
Attach it to the thing it should change, which is usually the scoping template, the qualification criteria, or the asset you reuse. A lesson filed as prose in a document is a nice sentiment. A lesson that edits the default effort assumption for a workstream, or adds a question to the intake form, or raises the risk score on a category of engagement, is a change to how the firm behaves. The app prompts for that step explicitly, asking which artifact the lesson should modify, because the gap between noticing something and changing something is where most improvement quietly dies.
