finups. Independent consultancy
Software optimization

Software efficiency for AI-heavy products

Better software.
Less waste.

I help AI-heavy SaaS and software teams reduce infrastructure, model/API, database, and workflow costs without compromising product quality. Every change is tested against the outcomes your users need.

Get a free snapshot Explore the work

Free Cost Opportunity Snapshot. One page, starting with 30 minutes with Fady.
No codebase access, API keys, or credentials needed.

Fewer detours. Same goal. Two routes connect the same starting point and destination. One takes several detours; the other goes directly between them. A conceptual illustration of reducing unnecessary work.
Fewer detours.
Same goal.
Efficiency is an engineering decision.Explore the evidence

How I can help

Find the waste.
Make the change count.

For AI-heavy software teams facing rising model bills, slow queries, or repeated processing. Infrastructure and workflow costs matter alongside model choice.

Free Cost Opportunity Snapshot

A one-page preliminary assessment, not a complete infrastructure audit.

  • One specific cost hypothesis
  • A benchmark or relevant portfolio example
  • Potential savings range with assumptions
  • Quality, reliability and implementation risks
  • Suggested next step

The savings range is a preliminary estimate, not verified savings for your product. No savings are guaranteed. No codebase access, API keys, or credentials needed. Further work starts only once we agree on scope and cost.

Optimization experiment

Test one change against the same workload, with quality and correctness checks agreed upfront.

You get A before-and-after comparison, documented tradeoffs, and a keep-or-reject recommendation.

Implementation & handover

Put a supported improvement into your codebase and make it clear how to verify and maintain it.

You get A focused code change, relevant checks, and rollout and rollback guidance.

The portfolio / Independent open-source work

Every saving has a story.
And a source.

Database queries, document processing, and AI workflows. See what changed, what the tests show, and where the limits are.

AnythingLLMMeasured locally

Database performance

A faster query.
An honest tradeoff.

Quota checks scanned chat records. A targeted database index made the tested busy-user query faster, with slower writes and extra storage as the tradeoff.

Before303.241 ms
After4.976 ms

Busy-user quota query · 1M rows · 8 MiB cache
Median of three per-trial medians

Technical details about AnythingLLM

Problem

Counting a user’s recent messages required scanning chat records. History queries needed those records filtered by conversation and returned in order. An index optimized only for the count could make history reads slower by introducing a separate sort.

Change

Use a composite index arranged around the user and conversation filters, followed by ordering and date fields. Quota counts can read the smaller index entries, while matching history queries can traverse them in order. The date still needs checking within the user’s entries.

Result & limitation

The tested large workloads with quota checks benefited. The index also adds about 9% storage and slows inserts. At a million rows, every tested sequence without the quota check was slower, making workload fit essential.

Evidence limits. Measured locally on synthetic data. Whole-app speed and bill savings not measured.

Experimental index · draft PR*Read the study
Paperless-GPTSynthetic test

Document intelligence

One document.
Fewer requests.

Five separate requests repeatedly processed the same document. Keeping the title first and grouping the remaining fields reduced that repetition in the synthetic test.

5requests before
2requests after
60% fewer application calls
Technical details about Paperless-GPT

Problem

Title, tags, correspondent, document type, and date were requested separately. That repeatedly sent the same document through the model. Later fields also depend on the generated title, so simply merging all five requests would lose useful context.

Change

Keep the title request, then ask for the other four fields together. Require a complete structured response and validate its fields. If grouping fails, return to the original individual requests. Custom prompts and other excluded configurations retain their existing behavior.

Result & limitation

The fixed synthetic example returned matching metadata with five requests reduced to two. A failed grouped request followed by successful fallback takes six, so the real benefit depends on how often grouping succeeds.

Evidence limits. Tested locally with simulated responses. Live accuracy and billed savings not measured.

Opt-in implementation · PR open*View the work

Opportunities under evaluation

What could improve next.

These estimates guide what to test. They are not measured savings: creative quality and successful caller outcomes still need live evaluation.

Open PomelliModeled estimate

Creative generation

Create with intention.
Reuse with care.

Reuse eligible creative assets, render simple motion locally, and track the cost of every attempt. A more deliberate workflow reduces the need to generate every variation from scratch.

45–55%modeled cost-reduction range

Eight-format campaign + two motion outputs

Technical details about Open Pomelli

Problem

Generating each platform format independently repeats image and copy work. Simple camera-like motion also goes through a generative-video service, even when it only needs a predictable transformation of an existing image.

Change

Separate reusable visual masters from platform exports. Use local rendering only for motion that does not need new action, geometry, or viewpoints. Track operations and attempts durably so retries, provider progress, and completion remain distinguishable.

Assumptions & limitations

Five reusable masters, one grouped copy request, 60–70% eligible local motion, and an allowance for local compute. Reuse still needs to preserve each format’s creative purpose. The broader implementation is local; the separate public PR fixes provider polling.

Evidence limits. Implementation checked locally. Savings are modeled; creative quality needs live evaluation.

Inspect the polling fix ↗
AIReceptionistModeled estimate

Realtime voice

A cheaper model.
The whole call matters.

Make a lower-cost voice model an explicit option, then evaluate it against the whole call: model usage, telephony, retries, and a successful caller outcome.

30–50%modeled whole-call savings

Depends on workload and comparable caller outcomes

Technical details about AIReceptionist

Problem

The realtime model is selected when a voice session is configured. Using a less expensive model can lower audio-processing costs, but telephony and session charges remain. Comparing model prices alone exaggerates the potential whole-call saving.

Change

Update the supported model options and make the smaller model an explicit configuration choice. Keep the full model as the default. Calculate both model and non-model costs, and evaluate transfers, interruptions, retries, and successful caller outcomes before deciding to switch.

Assumptions & limitations

The planning range uses a fixed speech workload and different non-model cost assumptions at September 1 prices. It depends on the smaller model preserving caller outcomes; the configuration change alone does not establish that.

Evidence limits. Configuration tested locally. Savings are modeled; live call quality not measured.

Inspect the implementation ↗

* Contribution statuses are a snapshot from 15 September 2026. All five upstream proposals were open and unmerged.

How I work

Less cost is useful.
A good outcome is essential.

The unit that matters is a successful user outcome. Failed attempts, retries, slow responses, and engineering effort belong in the calculation.

  1. 01

    Define what good means.

    Agree on the outcome and the quality that must be protected.

  2. 02

    Change one thing. Measure it.

    Compare the same workload and keep the evidence inspectable.

  3. 03

    Keep what earns its place.

    Record the benefit, the tradeoffs, and the reasons to keep or reject it.

The person behind the work

Work directly
with Fady.

I’m Fady, the person behind Finups, an independent software optimization consultancy. I investigate where software does unnecessary work, then test whether a focused change improves its cost or performance.

My open-source work spans database performance, document processing, and AI generation. You can inspect the changes, the tests, and the tradeoffs in each case.

Meet the work on GitHub

Start with one workflow

Where is your product
doing too much?

Start with a free 30-minute conversation with Fady. Get a one-page Cost Opportunity Snapshot: one cost hypothesis, a benchmark or relevant portfolio example, a potential savings range with assumptions, the risks, and a suggested next step. A preliminary assessment, not a complete infrastructure audit. No guaranteed savings or obligation to continue.

Get a free snapshot