DoneThat vs Pieces

The short answer

Pieces and DoneThat are the two products in this comparison that take AI-tool integration seriously. Both keep a record of your work and expose it through MCP so an assistant can use it.

The difference is what the record is for. Pieces builds a memory layer whose purpose is to make your assistants better. DoneThat builds a reconstruction of your working day whose purpose is to make you better, with MCP as one output among several. Pieces' integration breadth is genuinely stronger; DoneThat has coaching, goals and time reconstruction that Pieces does not attempt.

Where Pieces is genuinely ahead

MCP integration breadth. Unlimited MCP access to Claude, Cursor and other MCP-ready tools is the headline feature, and it is better than ours on this axis. If your goal is "every assistant I use should know what I have been working on", Pieces is built precisely for that.

Model flexibility. Claude, Gemini, ChatGPT and local models. You are not committed to one provider's roadmap or pricing.

On-device memory. The archive stays local, which is a stronger storage position than cloud-processed alternatives.

Cross-app capture rather than being trapped in one editor, plus one-click summaries and recaps.

A 14-day money-back guarantee on top of the trial, which is more generous than a trial alone.

Where the two diverge

Memory for assistants versus a record for you. Pieces' memory exists so your tools have context. It is infrastructure with a very good interface. DoneThat's record is the product — a reconstructed day you review, correct and set goals against.

Time. Pieces will not tell you that nine hours went to a recurring meeting this week. It is not measuring time; it is remembering content. If the question is "where did the week go", that is not what Pieces is for.

Coaching. Don nudges you onto goals, blocks calendar time for them, flags low-value work while you can still fix it, checks your mood, reminds you of meetings, suggests breaks. Pieces answers when asked.

Pricing

As of 11 August 2026:

Enterprise adds organisational control over approved providers and feature access. There is a 7-day trial that requires a credit card, plus a 14-day money-back guarantee.

Two notes. At $18.99 Pieces is at the expensive end of this comparison — you are paying for the integration depth. And the card-required trial is worth knowing about, since several competitors here (Toggl, Timing, ActivTrak, Rize) do not ask for one.

Platform support is not stated on the pricing page, so confirm your OS before subscribing.

DoneThat is subscription-based with a 14-day free trial; current prices are on the DoneThat download page.

On storage

Pieces keeps memory on-device. DoneThat never stores screenshots — processed in real time, immediately deleted, only the derived structured activity persists. Excluded apps are redacted on your machine before anything is sent, and with your own Gemini or OpenAI-compatible model raw activity never reaches our servers. The desktop app is open source.

Both are defensible. Pieces keeps a richer archive under your control; we keep a thinner one and give you the option of processing it entirely on your own model.

Where Pieces is the better choice

• AI assistant integration is the point. Unlimited MCP across clients is its strongest card.
• You work across several models and want to switch freely.
• You want on-device memory with a mature interface over it.
• Your work is code-heavy. The developer workflow focus shows, in a good way.
• You want organisational provider controls. Enterprise covers approved-provider governance.

Where DoneThat is the better choice

• You want to know where your time went. Pieces does not measure it.
• You want coaching, not recall. Nudges during the work rather than answers on request.
• You want goals tracked and something that notices drift.
• You want a team dimension — summary-level dashboards, relative trends rather than per-person hours, visibility set per summary, optional mood tracking.
• You want peer accountability. Follow others, share summaries, Slack and Discord, accountability groups — all opt-in.
• You want a trial without handing over a card.

Can you run both?

Plausibly, and they overlap less than the MCP-shaped marketing suggests: Pieces for content memory feeding your assistants, DoneThat for the time record and the coaching. Two subscriptions at these prices is a real cost, so this is for people who genuinely want both.

How we compared these

Everything stated about Pieces comes from its public pricing page, read on 11 August 2026 and linked below. Platform support is not quoted because the page does not state it.

Claims about DoneThat are first-hand: we build it. Pieces' MCP integration breadth is better than ours and its model flexibility is wider — both stated plainly above.

Questions people ask about Pieces

What does Pieces actually do?

It keeps an on-device long-term memory of your work across applications, then makes it available to AI tools — with unlimited MCP access to Claude, Cursor and other MCP-ready clients — plus one-click summaries and recaps. It is a memory layer for your assistants rather than a time tracker.

How much does Pieces cost?

Pro is $18.99 per month with a discounted yearly option; Enterprise is $22.99 per month or $69.99 per quarter and adds organisational controls. There is a 7-day trial that requires a credit card, and a 14-day money-back guarantee.

Is my data stored locally with Pieces?

Yes — memory is saved on-device, which is a stronger position than cloud-processed alternatives. DoneThat's approach differs: screenshots are never stored at all, and with your own model raw activity never reaches our servers.

Pieces vs DoneThat for MCP?

Both expose your history through MCP, and Pieces' integration breadth is better — unlimited access across Claude, Cursor and other clients is its headline feature. DoneThat's MCP server exposes reconstructed work history and adds goals and coaching on top. Pick Pieces if the assistant integration is the product; DoneThat if you also want the coaching.