SHIM

SHIM (domain getshim.tech) is an AI gateway: middleware that sits between an application and large language model providers such as OpenAI, Anthropic, and Google Gemini. It intercepts outbound API calls to redact personally identifiable information, serve semantically similar requests from a cache, and normalize model output, all through a single change to the API base URL.¹ The product is built by co-founders Can Cagatay Sevgican (product) and Mertcan Saglam (engineering), and is based in Istanbul, Turkey.²³ In May 2026, SHIM was selected as one of ten startups, from more than 800 applicants, for the 14th cohort of Workup, the startup program run by Turkiye Is Bankasi’s AI Factory.³

Product

SHIM is positioned as an enterprise gateway for AI traffic, governing the requests an application sends to LLM providers without requiring a rewrite of the application itself. Integration is a base-URL change rather than an SDK migration, so existing OpenAI-style calls route through SHIM unchanged.¹ Three functions sit at the core of the product:¹

• Semantic cache. SHIM fingerprints prompts using embeddings so that near-duplicate requests resolve to the same cached answer. Its own example is that “Recommend a movie” and “Suggest a film” hit the same cache entry, where an exact-match cache would not. SHIM reports up to 80 percent fewer redundant provider calls and roughly 40 percent lower OpenAI cost for repetitive workloads, both self-reported figures.¹
• PII redaction. Context-aware entity detection strips names, card numbers, secrets, and similar identifiers before a prompt leaves the customer’s infrastructure, then re-identifies the values locally in the response. SHIM frames this around GDPR and Turkey’s KVKK and references k-anonymity and l-diversity controls.¹
• Normalization. Unstructured model output is coerced into strict JSON schemas so downstream code receives a consistent shape on every call.¹

SHIM also exposes a usage dashboard covering request volume, estimated cost saved, cache-hit rate, PII events, and average latency over time.¹ The company states a middleware latency target under 100 milliseconds.¹

Pricing

SHIM’s published pricing, billed monthly with a roughly 20 percent discount on annual plans, as of June 2026:⁴

Tier	Monthly price	Requests / month	Aimed at
Developer	$0	3,000	Hobby projects and testing
Solo Pro	$29	100,000	Indie hackers and freelancers
Agency	$149	1,000,000	Software houses and agencies
Enterprise	Custom	Unlimited	On-premise, custom SLAs

The free Developer tier includes basic semantic cache and PII detection. Solo Pro adds the full PII shield, unlimited cache, priority support, and bring-your-own-key provider credentials. Agency adds multi-tenant management for up to ten clients, a white-label option, custom PII rules, and an onboarding call. Enterprise covers on-premise deployment and custom service-level agreements.⁴

Security and data handling

SHIM’s central claim is a zero-retention, pass-through architecture.¹ The company states that it uses Redis only for transient, automatically expiring cache entries and does not persist prompt or completion data to a permanent store or use it for training. Provider API keys are encrypted at rest and decrypted only in memory for the brief moment a request is proxied. Identity runs on Supabase Auth, which SHIM describes as “blind authentication” because passwords are hashed and never visible to the service. The architecture separates the control plane (customer settings) from the data plane (AI traffic).¹

Who it’s for

SHIM targets developers and teams running production AI where data privacy and cost control both matter, naming RAG pipelines, customer-support chatbots, and internal tools as typical use cases.¹ The tier structure maps cleanly onto three buyers: individual developers on the free Developer tier, indie hackers and freelancers on Solo Pro, and software houses or agencies that resell to multiple clients on the multi-tenant, white-label Agency tier.⁴

How it compares

SHIM sits in the AI gateway and LLM-proxy category, alongside both open-source self-hosted projects and managed platforms. The most frequently compared alternatives are below. All pricing is taken from each vendor’s own published pages, retrieved June 2026.

Product	Starting price	Hosting	Built-in PII redaction	Semantic cache
SHIM	Free; $29/mo (Solo Pro)4	Managed cloud1	Yes, from the free tier4	Yes, paid tiers1
Portkey	Free; $49/mo (Production)5	Managed or self-host5	At Enterprise tier5	At Enterprise tier5
Helicone	Free (10k req/mo); $79/mo (Pro)6	Managed or self-host6	Not specified6	Yes6
LiteLLM	Free (open source)7	Self-host7	Not built in7	Redis, self-configured8
Cloudflare AI Gateway	Free (all plans)9	Managed (edge)9	Not a core feature9	Yes (edge cache)9

SHIM is distinct in this set for offering PII redaction at its entry tiers rather than gating it to an enterprise contract, for a flat indie-friendly price (free, then $29 per month), and for a multi-tenant white-label Agency tier built for resellers.⁴ Portkey, the closest managed competitor, places semantic caching and PII anonymization at its Enterprise tier.⁵

Alternatives

The right choice depends on hosting preference and which guardrails matter most:

• Choose Portkey if you want the broadest managed guardrail and routing feature set and can move PII anonymization and semantic caching onto an enterprise contract.⁵
• Choose Helicone if your priority is observability and logging for LLM calls rather than a redaction-first gateway.⁶
• Choose LiteLLM if you want to self-host an open-source proxy across 100-plus providers and assemble caching and guardrails yourself.⁷
• Choose Cloudflare AI Gateway if you are already on Cloudflare and want edge caching, rate limiting, and fallback included across plans.⁹
• Choose SHIM if you want a managed gateway with PII redaction and semantic caching available from the free or $29 tier, a GDPR and KVKK framing, and a white-label option for reselling to clients.¹⁴

Verified performance

An independent Lighthouse audit of the getshim.tech homepage, run June 9, 2026, returned strong scores: Performance 93, SEO 100, Best Practices 100, and Accessibility 80 out of 100.¹⁰ The site publishes Organization, SoftwareApplication, and FAQPage structured data and maintains an active blog, documentation, and a playground under its sitemap.¹

Company

SHIM is built by two co-founders based in Istanbul:²

• Can Cagatay Sevgican, co-founder, product, covering AI engineering, full-stack development, and product strategy.²
• Mertcan Saglam, co-founder, engineering, covering data and NLP engineering and infrastructure.²

In May 2026, SHIM was named one of ten startups selected from more than 800 applicants for the 14th cohort of Workup, the entrepreneurship program run by Turkiye Is Bankasi through its AI Factory. Independent coverage described SHIM as an enterprise AI gateway that lets companies use AI models securely without compromising data privacy.³ The company foregrounds an environmental angle, arguing that cached requests avoid redundant computation and lower energy use, and it states alignment with the Coalition for Sustainable AI, the Green Software Foundation, and the SME Climate Hub.²

External links

• getshim.tech (official site)
• About and founders
• GitHub @getshim
• LinkedIn company page
• Workup Cohort 14 announcement (Webrazzi)

Footnotes

1. SHIM homepage, including positioning, the semantic-cache, PII-redaction, and normalization features, the zero-retention architecture, and embedded Organization, SoftwareApplication, and FAQPage schema. Retrieved June 9, 2026. https://getshim.tech/ ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵

2. SHIM about page (co-founders and roles, sustainability mission and stated alignments). Retrieved June 9, 2026. https://getshim.tech/about ↩ ↩² ↩³ ↩⁴ ↩⁵

3. “Workup Girisimcilik Programi’nin 14. donemine secilen 10 girisim,” Webrazzi, May 25, 2026. SHIM is named among the ten startups selected from more than 800 applications for Workup Cohort 14, run by Turkiye Is Bankasi. https://webrazzi.com/2026/05/25/workup-girisimcilik-programi-nin-14-donemine-secilen-10-girisim/ ↩ ↩² ↩³

4. SHIM pricing section (Developer, Solo Pro, Agency, and Enterprise tiers, request limits, and per-tier features). Retrieved June 9, 2026. https://getshim.tech/ ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷

5. Portkey pricing page (Developer free, Production $49/mo, Enterprise custom; semantic caching and PII anonymization at Enterprise). Retrieved June 9, 2026. https://portkey.ai/pricing ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶

6. Helicone pricing page (Hobby free at 10,000 requests/month, Pro $79/mo, Team $799/mo; observability focus). Retrieved June 9, 2026. https://www.helicone.ai/pricing ↩ ↩² ↩³ ↩⁴ ↩⁵

7. LiteLLM site (open-source self-hosted proxy across 100-plus providers, free; Enterprise custom). Retrieved June 9, 2026. https://www.litellm.ai/ ↩ ↩² ↩³ ↩⁴

8. AI gateway comparison roundups used for cross-vendor feature positioning (TECHSY, “8 Gateways Ranked 2026”; Deepak Gupta, “Top 5 AI Gateways 2026”). Retrieved June 9, 2026. ↩

9. Cloudflare AI Gateway documentation (available on all plans; caching, rate limiting, fallback, observability). Retrieved June 9, 2026. https://developers.cloudflare.com/ai-gateway/ ↩ ↩² ↩³ ↩⁴ ↩⁵

10. SHIM Lighthouse audit of the homepage (Performance 93, Accessibility 80, Best Practices 100, SEO 100). Run June 9, 2026. ↩