fix-summary-responsive.review.docs.asknews.dev Open in urlscan Pro
23.173.114.114  Public Scan

Form analysis
0 forms found in the DOM

Text Content

 * Home
 * API Reference


Search documentation...⌘K

Search documentation...⌘K

 * 





 * 🏠
   Getting Started
 * 📰
   News
 * 📖
   Stories
 * 💬
   Chat
 * 🔮
   Forecast
 * 👄
   Reddit
 * 📈
   Analytics
 * 🔎
   Sources
 * 🔗
   LangChain

#
OVERVIEW

The AskNews SDK allows you to access a variety of news related endpoints. This
includes:

 * 📰 News: Natural language, low-latency, prompt-optimized, endpoint. Ready to
   be the news-context link in your LLM chain.
 * 📖 Stories: Get the hottest topics currently circulating in the news-sphere.
   Track sentiment, coverage, source origin, through time. Benefit from
   state-of-the-art clustering and custom written stories based on
   human-in-the-loop editorial.
 * 💬 Chat: News infused assistant, interact with it like any other OpenAI
   model.
 * 🔮 Forecast: Make any news-related forecast imaginable, based on rich
   temporal context and SOTA LLMs.
 * 👄 Reddit: Search Reddit live, get fully structured and GPT-summarized
   results - ready to contextualize your LLM.
 * 📈 Analytics: AskNews analytics on financial assets and politics.
 * 🔎 Sources: Review the distribution of sources underpinning AskNews for any
   given period of time.
 * 🔗 LangChain: Use AskNews as a LangChain retriever or tool.

AskNews infuses any LLM with the latest news, using a single natural language
query. Specifically, AskNews is enriching over 300k articles per day, indexing
them into a hot vector database, and putting that vector database on an endpoint
for you. When you query AskNews, you get back a prompt-optimized string that
goes directly into your prompt. This means that you do not need to manage your
own news RAG, and you do not need to worry about how to properly convey news
information in a condensed way to your LLM.

Python (sync)TypeScript
1from asknews_sdk import AskNewsSDK 2 3ask = AskNewsSDK() 4query = "Effect of
fed policy on tech sector" 5 6# prompt-optimized string ready to go for any LLM:
7news_context = ask.news.search_news(query).as_string

#
WHAT IS ASKNEWS DOING TO KEEP DATA HOT AND HIGH-QUALITY?

Building and managing a high-quality real-time news Retrieval Augmented
Generation (RAG) architecture includes:

 * 🪣 Scraping 50k news websites every 5 minutes,
 * 🧽 Cleaning, summarizing, translating, and enriching 300k articles per day,
 * 🧮 Embedding the articles with dense and sparse vectors,
 * 💾 Storing the documents in an ever-growing vector database,
 * 🔬 Monitoring for quality and ensuring up-time reliability,
 * 🏎 Ensuring low-latency interactions to avoid slowing down your LLM
   application,
 * 🧑‍🔬️ Researching and developing state of the art (SOTA) methods for entity
   extraction. and quality/accuracy control,
 * 🧭 Researching and improving methods for retrieval on dense and sparse vector
   indices,
 * 🌍 Tracking news narratives through time with SOTA clustering/tracking
   methods.

When you use AskNews, you get all this with a single line of code.

#
PERFORMANCE 🏎

A quantitative benchmark was run on AskNews vs JinaAI vs Tavily vs Exa, results
indicate that AskNews is 1400% faster than JinaAI, and has 78% better context
precision than all competitors. The full blog + Google Colab is available here.
The summary of results is shown here:



#
RESEARCH 🧑🏽‍🔬️

We researched, developed, and deployed the best entity extraction model in the
world, called GLiNER News. It is currently used to extract entities for all
articles and stories in the AskNews database. It is also completely open-source,
in-case you'd like to run the entity extraction yourself. You can find it on our
HuggingFace repository.

Our research extends beyond GLiNER, we are actively improving quality and
innovating on all aspects of our system. Our full list of scientific
publications can be found here.

#
QUALITY GUARANTEE 👑

This level of dedication to quality and transparency sets us apart from all
other News APIs. When you consume AskNews content, you are getting only the best
quality data, guaranteed. We even have our own Quality Assurance and Quality
Control that runs on every single article/story to ensure that you will not run
into a single article with incorrect content, scraping, etc.

#
TRANSPARENCY 🔬

Much of this quality comes from our dedication to high-quality foundational
software. We also developed and open-sourced Flowdapt, which is a cluster
orchestration software that enables highly reactive and adaptive artificial
intelligence software. This software is what powers the AskNews backend,
ensuring that we can keep up with the ever-changing news landscape, without
missing a single article.

Beyond these points, we are dedicated to open-source, with the majority of our
software completely free and open-source. We are fully transparent about our
datasources as well, with a transparency dashboard available to track and
monitor the data. We are commonly invited to present on our open-source
methodologies at conferences, such as GenAI Zurich. You can find all our
presentations here.

#
SUPPORT

If you need help with the AskNews SDK, please join our Discord, where you can
ask questions, share your projects, and connect with other AskNews users.

If you would like to explore the data before using the SDK, we encourage you to
visit the AskNews website at https://asknews.app.

#
ACADEMIC USE

We love supporting academic use-cases. If you are interested in accessing
AskNews data for an academic project, please reach out to us at
contact@asknews.app with a description of your project as well as contact
information. In most cases, academic access to AskNews is completely free under
the Limited License.

#
QUICKSTART

#
INSTALL

We are going to setup an example that infuses GPT3.5 with the latest news, but
you could substitute GPT3.5 with any LLM you like. First, install the python
package:

pip install asknews


or install the TypeScript SDK:

npm install @emergentmethods/asknews-typescript-sdk


#
QUICK NEWS CONTEXT IN YOUR LLM

Next, import the AskNews and OpenAI SDKs:

Python (sync)Python (async)TypeScript
1from asknews_sdk import AskNewsSDK 2from openai import OpenAI 3 4ask =
AskNewsSDK( 5 client_id="your_client_id", 6 client_secret="your_client_secret",
7 scopes=["chat", "news", "stories"] 8) 9oai_client = OpenAI(api_key="")



Your AskNews client_id and client_secret can be generated by going to the
AskNews console and creating an account. You create your API credentials and set
them in the AskNewsSDK constructor above.

Let's take an example of your LLM interacting with a user who is asking about
the current political situation in Germany. If you had your own RAG setup, you
would use that query to go search a database and retrieve documents, structure
them into a prompt-ready format, and it inject into your prompt. With AskNews,
you treat it as a one-stop RAG shop. Here's how you would do it:

Python (sync)Python (async)TypeScript
1def main(): 2 # Your user asks a question about the current political situation
in Germany 3 user = { 4 "role": "user", 5 "content": "What is the current
political situation in Germany?" 6 } 7 8 # Grab a prompt-optimized string ready
to go for your LLM: 9 response = ask.news.search_news( 10 query=user["content"],
# any natural language query 11 n_articles=10, # control the number of articles
to include in the context 12 return_type="string", # you can also ask for
"dicts" if you want more information 13 method="nl" # use "nl" for natural
language for your search, or "kw" for keyword search 14 ) 15 # now you have a
prompt optimized string: 16 news_articles = response.as_string 17 18 # simply
infuse that string into the prompt: 19 system = { 20 "role": "system", 21
"content": "A chat between a curious user and an artificial intelligence
Assistant. " 22 "The Assistant has access to the following news articles that "
23 f"may be useful for answering the User's questions: {news_articles}" 24 } 25
response = oai_client.chat.completions.create( 26 model="gpt-3.5-turbo", 27
messages=[system, user] 28 ) 29 30 print(response.choices[0].message.content) 31
32main()



Now the response provides the latest updated information about the political
situation in Germany:

--------------------------------------------------------------------------------

Response from GPT3.5

The current political situation in Germany is dynamic and multifaceted.

 1. Internal Security Concerns: There are doubts about Germany's internal
    security, with increasing threats from outside and vulnerabilities within
    the state structure, as highlighted in debates over strengthening the
    Federal Constitutional Court.

 2. Fiscal Policy: The CDU's party leader, Friedrich Merz, has raised concerns
    about the financial feasibility of spending 40 billion euros on citizen's
    income in 2024, especially in the context of maintaining defense
    capabilities.

 3. Political Shifts: The SPD has been criticized for moving away from its core
    voter base and reviving old Russia policies. This has sparked warnings about
    denying reality and concerns about Germany's foreign policy post-Merkel.

 4. Youth Political Views: A survey found that a significant portion of
    first-time voters lean towards conservative views, some even trusting the
    far-right Alternative for Germany (AfD)...(LLM response truncated for
    brevity)

--------------------------------------------------------------------------------

More details about filtering and sorting your news search can be found here.

#
GETTING THE HOTTEST TOPICS IN THE NEWS-SPHERE

If you would like to obtain the hottest tech stories in North America, you can
use SDK as follows:

Python (sync)Python (async)TypeScript
1def get_stories(): 2 response = ask.stories.search_stories( 3
categories=["Technology", "Science"], 4 continent="North America", 5
sort_by="coverage", # sort by coverage 6 sort_type="desc", # highest to lowest
coverage 7 reddit=3, # include 3 reddit threads 8 expand_updates=True, # get all
the details for updates 9 max_updates=2, # get the 2 most recent updates for
each story 10 max_articles=10 # include 10 news articles associated with each
update 11 ) 12 print([story.updates[0].headline for story in response.stories])
13 14get_stories()



Which would return the hottest tech stories in North America, sorted by
coverage, with the top 3 Reddit threads, and the 2 most recent updates for each
story. Each story in the response is a custom AskNews-written story about a
topic that is currently circulating in the news-sphere. The story includes a
host of aggregated information such as sentiment, reddit opinion, clustered
articles, coverage counts, origin diversity and much more.

More details about filtering and sorting your stories search can be found here.

#
USE THE NEWS-INFUSED CHAT ENDPOINT

If you would like to use the chat endpoint to ask questions about the news, you
can use the SDK as follows:

Python (sync)Python (async)TypeScript
1def chat_query(): 2 response = ask.chat.get_chat_completions( 3 messages=[ 4 {
5 "role": "user", 6 "content": "What is the top tech news?" 7 } 8 ], 9
stream=False 10 ) 11 12 # response object maches the OpenAI SDK response object:
13 print(response.choices[0].message.content) 14 15chat_query()



Beyond news, there are other endpoints like stories, chat, analytics, and
sources that you can explore.

#
ADDITIONAL LEARNING MATERIAL

Check out our blog post explaining how to use the AskNews SDK to infuse news
into your LLM.


TABLE OF CONTENTS

 * What is AskNews doing to keep data hot and high-quality?
   * Performance 🏎
   * Research 🧑🏽‍🔬️
   * Quality Guarantee 👑
   * Transparency 🔬
 * Support
 * Academic use
 * Quickstart
   * Install
   * Quick News context in your LLM
   * Getting the hottest topics in the news-sphere
   * Use the News-infused Chat endpoint
   * Additional learning material

Terms of ServicePrivacy PolicyCookie Preferences


Made with ❤️ by © Emergent Methods, LLC

Made with ❤️ by © Emergent Methods, LLC

Terms of ServicePrivacy PolicyCookie Preferences



WE USE COOKIES

We use cookies to ensure you get the best experience on our website. Some of
these cookies are provided by third parties. You are free to decide which
categories you would like to permit and can withdraw this consent at any time
(via cookie preferences link on the footer).

You can either accept all cookies, reject all but the necessary cookies or click
the "Preferences" button to decide which cookie categories you would like to
enable or disable.
Learn more on our privacy policy page.

PreferencesOnly NecessaryAccept All