RIP, Harvest

A popular time-tracking and invoicing service, Harvest, has sharply raised prices and added complex usage-based and revenue-skimming fees after its acquisition by Bending Spoons, prompting long-time customers to leave. Many small consultancies and freelancers describe feeling gouged and locked in, yet increasingly conclude it’s viable to replace such tools with lightweight in-house apps or AI-assisted “vibe coded” clones tailored to their needs. The episode is cited as evidence that aggressive post-acquisition monetization can rapidly erode trust and that AI is weakening the moat around traditional SaaS products, especially simple billing, invoicing, and workflow tools.

Harvest price hikes & confusing new model

  • Multiple long‑time users report bills jumping 8–10x or more (e.g., ~$40→$586/mo, ~$1500/yr→$20k/yr, ~$150→$1100+).
  • New “Flex” / usage‑based pricing appears to:
    • Charge per seat plus for projects/clients/tasks.
    • Skim a percentage of invoice value, even when Harvest doesn’t process payments (e.g., ACH/wire).
  • Some commenters initially doubt the math based on public pricing pages; others link support docs and personal emails showing large jumps once usage fees are added.
  • Consensus: pricing page is opaque; “starting from” numbers hide real costs and can produce shock renewals.

Customer reactions & trust

  • Many users cancel after the increase; some execute chargebacks, others downsize to minimum seats or move to monthly billing to limit surprise.
  • Several say the product had been “just fine” and reasonably priced for a decade; the trust damage from sudden gouging is described as permanent.
  • Some see this as a textbook way to drive users to competitors or DIY solutions.

DIY and AI‑assisted replacements

  • Numerous commenters “vibe coded” internal replacements for time tracking/invoicing, ranging from a weekend hack to ~10 days of work.
  • LLMs are credited with:
    • Quickly scaffolding CRUD webapps around SQLite/Beancount/etc.
    • Importing old data and generating invoices from existing exports.
  • Skeptics argue:
    • You can’t safely “one‑shot prompt” complex or financial systems.
    • LLM‑written code can be sloppy/buggy; real productization and maintenance still take time.
  • Common pattern: small tech orgs build narrowly tailored tools, then mostly stop adding features once needs are met.

SaaS moats in the AI era

  • Many see time tracking/invoicing and lightweight project tools as having “no moat” now that AI can accelerate clones.
  • Others note that:
    • Maintenance burden, edge cases, and integrations still matter.
    • Non‑technical or large organizations face higher switching costs, so some will remain despite price hikes.

Views on Bending Spoons / private equity model

  • Widespread perception of a playbook: acquire mature SaaS, slash staff, replace support with AI, drastically raise prices, accept churn, keep high‑margin survivors.
  • Several examples cited beyond Harvest (e.g., Evernote, Vimeo, MileIQ) with similar narratives.
  • Debate over causes: some argue only struggling or investor‑pressured businesses sell; others blame founder burnout and lack of gentler exit options.