Research at DailyFit
Self-accelerating Agentic AI.
The next problem we choose to solve.
Science advances by hypothesis and experiment.
AI is starting to run that cycle on its own.
That is when progress outgrows human speed.
DailyFit is bringing that moment closer.
Self-acceleration · an Agent that learns and improves on its own
An AI-native company
A product company.
And an AI research company.
The next era of software is the Agent that works on its own.
DailyFit proves that principle in a real, running service.
Research pushes the product, and the product proves the research.
Proven live
Validated in a live service, not on a benchmark.
Loops pay rent
Every learning loop must prove its own cost.
Users decide
The Agent proposes; the user always makes the call.
Reversible
Every intervention is logged and instantly reversible.
Core research theme
An Agent that evolves
before it’s told to.
The Agent repeats the same procedures every day and recognizes the patterns in its obstacles.
That learning is applied to the next repetition, by the Agent itself.
The goal is a single state.
Before anyone asks for an improvement, the Agent is already better.
ordinary automation
self-accelerating Agent
The result changes the next run.
That is why it gets faster with use.
The method
Acceleration is built in five stages.
Not vague self-improvement: a measurable pipeline.
Every stage is logged, measured, and gated.
- 01observe
Observe
Every run is logged as a structured trace.
- 02hypothesize
Hypothesize
The Agent forms improvement hypotheses from failure patterns.
- 03intervene
Intervene
Strategies change, inside guardrails.
- 04measure
Measure
Performance deltas are quantified against prior runs.
- 05consolidate
Consolidate
Only validated learning is written to long-term memory.
Why DailyFit is the testbed
A real service, running daily,
is the best laboratory.
The Auto-apply Agent collides with real portals, forms, and procedures every day.
Those repetitions and failures become the learning data.
The proving ground is not a paper benchmark.
It is a living service.
run #847 · auto-apply · community portal
obstacleObstacleform changed · first attempt failed
learnPattern storedmatch by label, not position · strategy updated
applySelf-appliedcarried into the next run, unprompted
run #848same portal · passes without retry
- processing time Δ -14s
- human input: zero
Open questions
The questions we haven’t solved
When should an Agent learn
Always-on learning, or check-ins on a fixed cycle.
The timing of learning is itself a design problem.
How much is too much
Over-learning disturbs the flow and reinforces the wrong directions.
We assume an optimal frequency exists, and we search for it.
What does acceleration cost
Self-acceleration that burns unlimited tokens can improve less than it spends.
ROI gates every learning loop.
We’re looking for the people who want to solve them.
Research frontiers
One principle,
expanding into every domain.
Self-acceleration is only the beginning.
The same principle extends beyond hobbies into jobs, beyond personal life into professional life.
And beyond finding opportunities, into creating them.
From hobbies to jobs
The principle that designs a day applies unchanged to finding work opportunities.
From personal life to professional life, one Agent carries it all.
From finding to creating
Finding opportunities is not the finish line.
We study the stage where the Agent creates activities and jobs on its own.
The next topics are in preparation
More research is already lined up on the long-term roadmap.
Each goes public once it clears validation.